intersect: utils/gff_util.py annotate

annotate utils/gff_util.py @ 5:60925436ca5f draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689

author	devteam
date	Thu, 22 Jun 2017 18:50:20 -0400
parents
children

rev	line source
5 60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	1 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	2 Provides utilities for working with GFF files.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	3 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	4 import copy
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	5
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	6 from bx.intervals.io import GenomicInterval, GenomicIntervalReader, MissingFieldError, NiceReaderWrapper
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	7 from bx.tabular.io import Comment, Header, ParseError
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	8
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	9 from .odict import odict
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	10
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	11
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	12 class GFFInterval( GenomicInterval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	13 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	14 A GFF interval, including attributes. If file is strictly a GFF file,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	15 only attribute is 'group.'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	16 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	17 def __init__( self, reader, fields, chrom_col=0, feature_col=2, start_col=3, end_col=4,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	18 strand_col=6, score_col=5, default_strand='.', fix_strand=False ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	19 # HACK: GFF format allows '.' for strand but GenomicInterval does not. To get around this,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	20 # temporarily set strand and then unset after initing GenomicInterval.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	21 unknown_strand = False
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	22 if not fix_strand and fields[ strand_col ] == '.':
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	23 unknown_strand = True
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	24 fields[ strand_col ] = '+'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	25 GenomicInterval.__init__( self, reader, fields, chrom_col, start_col, end_col, strand_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	26 default_strand, fix_strand=fix_strand )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	27 if unknown_strand:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	28 self.strand = '.'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	29 self.fields[ strand_col ] = '.'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	30
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	31 # Handle feature, score column.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	32 self.feature_col = feature_col
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	33 if self.feature_col >= self.nfields:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	34 raise MissingFieldError( "No field for feature_col (%d)" % feature_col )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	35 self.feature = self.fields[ self.feature_col ]
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	36 self.score_col = score_col
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	37 if self.score_col >= self.nfields:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	38 raise MissingFieldError( "No field for score_col (%d)" % score_col )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	39 self.score = self.fields[ self.score_col ]
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	40
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	41 # GFF attributes.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	42 self.attributes = parse_gff_attributes( fields[8] )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	43
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	44 def copy( self ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	45 return GFFInterval(self.reader, list( self.fields ), self.chrom_col, self.feature_col, self.start_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	46 self.end_col, self.strand_col, self.score_col, self.strand)
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	47
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	48
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	49 class GFFFeature( GFFInterval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	50 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	51 A GFF feature, which can include multiple intervals.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	52 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	53 def __init__( self, reader, chrom_col=0, feature_col=2, start_col=3, end_col=4,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	54 strand_col=6, score_col=5, default_strand='.', fix_strand=False, intervals=[],
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	55 raw_size=0 ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	56 # Use copy so that first interval and feature do not share fields.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	57 GFFInterval.__init__( self, reader, copy.deepcopy( intervals[0].fields ), chrom_col, feature_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	58 start_col, end_col, strand_col, score_col, default_strand,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	59 fix_strand=fix_strand )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	60 self.intervals = intervals
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	61 self.raw_size = raw_size
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	62 # Use intervals to set feature attributes.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	63 for interval in self.intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	64 # Error checking. NOTE: intervals need not share the same strand.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	65 if interval.chrom != self.chrom:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	66 raise ValueError( "interval chrom does not match self chrom: %s != %s" %
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	67 ( interval.chrom, self.chrom ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	68 # Set start, end of interval.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	69 if interval.start < self.start:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	70 self.start = interval.start
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	71 if interval.end > self.end:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	72 self.end = interval.end
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	73
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	74 def name( self ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	75 """ Returns feature's name. """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	76 name = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	77 # Preference for name: GTF, GFF3, GFF.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	78 for attr_name in ['gene_id', 'transcript_id', # GTF
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	79 'ID', 'id', # GFF3
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	80 'group' ]: # GFF (TODO)
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	81 name = self.attributes.get( attr_name, None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	82 if name is not None:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	83 break
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	84 return name
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	85
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	86 def copy( self ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	87 intervals_copy = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	88 for interval in self.intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	89 intervals_copy.append( interval.copy() )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	90 return GFFFeature(self.reader, self.chrom_col, self.feature_col, self.start_col, self.end_col, self.strand_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	91 self.score_col, self.strand, intervals=intervals_copy )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	92
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	93 def lines( self ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	94 lines = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	95 for interval in self.intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	96 lines.append( '\t'.join( interval.fields ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	97 return lines
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	98
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	99
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	100 class GFFIntervalToBEDReaderWrapper( NiceReaderWrapper ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	101 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	102 Reader wrapper that reads GFF intervals/lines and automatically converts
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	103 them to BED format.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	104 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	105
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	106 def parse_row( self, line ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	107 # HACK: this should return a GFF interval, but bx-python operations
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	108 # require GenomicInterval objects and subclasses will not work.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	109 interval = GenomicInterval( self, line.split( "\t" ), self.chrom_col, self.start_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	110 self.end_col, self.strand_col, self.default_strand,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	111 fix_strand=self.fix_strand )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	112 interval = convert_gff_coords_to_bed( interval )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	113 return interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	114
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	115
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	116 class GFFReaderWrapper( NiceReaderWrapper ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	117 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	118 Reader wrapper for GFF files.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	119
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	120 Wrapper has two major functions:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	121
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	122 1. group entries for GFF file (via group column), GFF3 (via id attribute),
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	123 or GTF (via gene_id/transcript id);
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	124 2. convert coordinates from GFF format--starting and ending coordinates
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	125 are 1-based, closed--to the 'traditional'/BED interval format--0 based,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	126 half-open. This is useful when using GFF files as inputs to tools that
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	127 expect traditional interval format.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	128 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	129
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	130 def __init__( self, reader, chrom_col=0, feature_col=2, start_col=3,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	131 end_col=4, strand_col=6, score_col=5, fix_strand=False, convert_to_bed_coord=False, **kwargs ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	132 NiceReaderWrapper.__init__( self, reader, chrom_col=chrom_col, start_col=start_col, end_col=end_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	133 strand_col=strand_col, fix_strand=fix_strand, **kwargs )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	134 self.feature_col = feature_col
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	135 self.score_col = score_col
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	136 self.convert_to_bed_coord = convert_to_bed_coord
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	137 self.last_line = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	138 self.cur_offset = 0
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	139 self.seed_interval = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	140 self.seed_interval_line_len = 0
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	141
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	142 def parse_row( self, line ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	143 interval = GFFInterval( self, line.split( "\t" ), self.chrom_col, self.feature_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	144 self.start_col, self.end_col, self.strand_col, self.score_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	145 self.default_strand, fix_strand=self.fix_strand )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	146 return interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	147
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	148 def __next__( self ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	149 """ Returns next GFFFeature. """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	150
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	151 #
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	152 # Helper function.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	153 #
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	154
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	155 def handle_parse_error( parse_error ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	156 """ Actions to take when ParseError found. """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	157 if self.outstream:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	158 if self.print_delegate and hasattr(self.print_delegate, "__call__"):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	159 self.print_delegate( self.outstream, e, self )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	160 self.skipped += 1
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	161 # no reason to stuff an entire bad file into memmory
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	162 if self.skipped < 10:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	163 self.skipped_lines.append( ( self.linenum, self.current_line, str( e ) ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	164
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	165 # For debugging, uncomment this to propogate parsing exceptions up.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	166 # I.e. the underlying reason for an unexpected StopIteration exception
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	167 # can be found by uncommenting this.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	168 # raise e
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	169
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	170 #
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	171 # Get next GFFFeature
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	172 #
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	173 raw_size = self.seed_interval_line_len
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	174
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	175 # If there is no seed interval, set one. Also, if there are no more
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	176 # intervals to read, this is where iterator dies.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	177 if not self.seed_interval:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	178 while not self.seed_interval:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	179 try:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	180 self.seed_interval = GenomicIntervalReader.next( self )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	181 except ParseError as e:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	182 handle_parse_error( e )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	183 # TODO: When no longer supporting python 2.4 use finally:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	184 # finally:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	185 raw_size += len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	186
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	187 # If header or comment, clear seed interval and return it with its size.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	188 if isinstance( self.seed_interval, ( Header, Comment ) ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	189 return_val = self.seed_interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	190 return_val.raw_size = len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	191 self.seed_interval = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	192 self.seed_interval_line_len = 0
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	193 return return_val
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	194
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	195 # Initialize feature identifier from seed.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	196 feature_group = self.seed_interval.attributes.get( 'group', None ) # For GFF
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	197 # For GFF3
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	198 feature_id = self.seed_interval.attributes.get( 'ID', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	199 # For GTF.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	200 feature_transcript_id = self.seed_interval.attributes.get( 'transcript_id', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	201
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	202 # Read all intervals associated with seed.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	203 feature_intervals = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	204 feature_intervals.append( self.seed_interval )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	205 while True:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	206 try:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	207 interval = GenomicIntervalReader.next( self )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	208 raw_size += len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	209 except StopIteration as e:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	210 # No more intervals to read, but last feature needs to be
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	211 # returned.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	212 interval = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	213 raw_size += len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	214 break
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	215 except ParseError as e:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	216 handle_parse_error( e )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	217 raw_size += len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	218 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	219 # TODO: When no longer supporting python 2.4 use finally:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	220 # finally:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	221 # raw_size += len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	222
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	223 # Ignore comments.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	224 if isinstance( interval, Comment ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	225 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	226
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	227 # Determine if interval is part of feature.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	228 part_of = False
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	229 group = interval.attributes.get( 'group', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	230 # GFF test:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	231 if group and feature_group == group:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	232 part_of = True
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	233 # GFF3 test:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	234 parent_id = interval.attributes.get( 'Parent', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	235 cur_id = interval.attributes.get( 'ID', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	236 if ( cur_id and cur_id == feature_id ) or ( parent_id and parent_id == feature_id ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	237 part_of = True
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	238 # GTF test:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	239 transcript_id = interval.attributes.get( 'transcript_id', None )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	240 if transcript_id and transcript_id == feature_transcript_id:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	241 part_of = True
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	242
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	243 # If interval is not part of feature, clean up and break.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	244 if not part_of:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	245 # Adjust raw size because current line is not part of feature.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	246 raw_size -= len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	247 break
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	248
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	249 # Interval associated with feature.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	250 feature_intervals.append( interval )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	251
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	252 # Last interval read is the seed for the next interval.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	253 self.seed_interval = interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	254 self.seed_interval_line_len = len( self.current_line )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	255
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	256 # Return feature.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	257 feature = GFFFeature( self, self.chrom_col, self.feature_col, self.start_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	258 self.end_col, self.strand_col, self.score_col,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	259 self.default_strand, fix_strand=self.fix_strand,
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	260 intervals=feature_intervals, raw_size=raw_size )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	261
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	262 # Convert to BED coords?
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	263 if self.convert_to_bed_coord:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	264 convert_gff_coords_to_bed( feature )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	265
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	266 return feature
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	267 next = __next__ # This line should be removed once the bx-python port to Python3 is finished
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	268
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	269
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	270 def convert_bed_coords_to_gff( interval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	271 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	272 Converts an interval object's coordinates from BED format to GFF format.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	273 Accepted object types include GenomicInterval and list (where the first
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	274 element in the list is the interval's start, and the second element is
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	275 the interval's end).
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	276 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	277 if isinstance( interval, GenomicInterval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	278 interval.start += 1
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	279 if isinstance( interval, GFFFeature ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	280 for subinterval in interval.intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	281 convert_bed_coords_to_gff( subinterval )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	282 elif type( interval ) is list:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	283 interval[ 0 ] += 1
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	284 return interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	285
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	286
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	287 def convert_gff_coords_to_bed( interval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	288 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	289 Converts an interval object's coordinates from GFF format to BED format.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	290 Accepted object types include GFFFeature, GenomicInterval, and list (where
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	291 the first element in the list is the interval's start, and the second
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	292 element is the interval's end).
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	293 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	294 if isinstance( interval, GenomicInterval ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	295 interval.start -= 1
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	296 if isinstance( interval, GFFFeature ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	297 for subinterval in interval.intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	298 convert_gff_coords_to_bed( subinterval )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	299 elif type( interval ) is list:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	300 interval[ 0 ] -= 1
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	301 return interval
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	302
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	303
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	304 def parse_gff_attributes( attr_str ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	305 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	306 Parses a GFF/GTF attribute string and returns a dictionary of name-value
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	307 pairs. The general format for a GFF3 attributes string is
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	308
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	309 name1=value1;name2=value2
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	310
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	311 The general format for a GTF attribute string is
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	312
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	313 name1 "value1" ; name2 "value2"
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	314
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	315 The general format for a GFF attribute string is a single string that
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	316 denotes the interval's group; in this case, method returns a dictionary
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	317 with a single key-value pair, and key name is 'group'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	318 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	319 attributes_list = attr_str.split(";")
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	320 attributes = {}
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	321 for name_value_pair in attributes_list:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	322 # Try splitting by '=' (GFF3) first because spaces are allowed in GFF3
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	323 # attribute; next, try double quotes for GTF.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	324 pair = name_value_pair.strip().split("=")
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	325 if len( pair ) == 1:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	326 pair = name_value_pair.strip().split("\"")
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	327 if len( pair ) == 1:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	328 # Could not split for some reason -- raise exception?
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	329 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	330 if pair == '':
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	331 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	332 name = pair[0].strip()
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	333 if name == '':
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	334 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	335 # Need to strip double quote from values
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	336 value = pair[1].strip(" \"")
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	337 attributes[ name ] = value
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	338
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	339 if len( attributes ) == 0:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	340 # Could not split attributes string, so entire string must be
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	341 # 'group' attribute. This is the case for strictly GFF files.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	342 attributes['group'] = attr_str
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	343 return attributes
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	344
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	345
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	346 def gff_attributes_to_str( attrs, gff_format ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	347 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	348 Convert GFF attributes to string. Supported formats are GFF3, GTF.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	349 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	350 if gff_format == 'GTF':
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	351 format_string = '%s "%s"'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	352 # Convert group (GFF) and ID, parent (GFF3) attributes to transcript_id, gene_id
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	353 id_attr = None
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	354 if 'group' in attrs:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	355 id_attr = 'group'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	356 elif 'ID' in attrs:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	357 id_attr = 'ID'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	358 elif 'Parent' in attrs:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	359 id_attr = 'Parent'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	360 if id_attr:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	361 attrs['transcript_id'] = attrs['gene_id'] = attrs[id_attr]
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	362 elif gff_format == 'GFF3':
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	363 format_string = '%s=%s'
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	364 attrs_strs = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	365 for name, value in attrs.items():
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	366 attrs_strs.append( format_string % ( name, value ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	367 return " ; ".join( attrs_strs )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	368
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	369
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	370 def read_unordered_gtf( iterator, strict=False ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	371 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	372 Returns GTF features found in an iterator. GTF lines need not be ordered
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	373 or clustered for reader to work. Reader returns GFFFeature objects sorted
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	374 by transcript_id, chrom, and start position.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	375 """
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	376
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	377 # -- Get function that generates line/feature key. --
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	378
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	379 def get_transcript_id(fields):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	380 return parse_gff_attributes( fields[8] )[ 'transcript_id' ]
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	381
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	382 if strict:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	383 # Strict GTF parsing uses transcript_id only to group lines into feature.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	384 key_fn = get_transcript_id
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	385 else:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	386 # Use lenient parsing where chromosome + transcript_id is the key. This allows
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	387 # transcripts with same ID on different chromosomes; this occurs in some popular
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	388 # datasources, such as RefGenes in UCSC.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	389 def key_fn(fields):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	390 return fields[0] + '_' + get_transcript_id( fields )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	391
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	392 # Aggregate intervals by transcript_id and collect comments.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	393 feature_intervals = odict()
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	394 comments = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	395 for count, line in enumerate( iterator ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	396 if line.startswith( '#' ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	397 comments.append( Comment( line ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	398 continue
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	399
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	400 line_key = key_fn( line.split('\t') )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	401 if line_key in feature_intervals:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	402 feature = feature_intervals[ line_key ]
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	403 else:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	404 feature = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	405 feature_intervals[ line_key ] = feature
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	406 feature.append( GFFInterval( None, line.split( '\t' ) ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	407
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	408 # Create features.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	409 chroms_features = {}
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	410 for count, intervals in enumerate( feature_intervals.values() ):
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	411 # Sort intervals by start position.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	412 intervals.sort( lambda a, b: cmp( a.start, b.start ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	413 feature = GFFFeature( None, intervals=intervals )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	414 if feature.chrom not in chroms_features:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	415 chroms_features[ feature.chrom ] = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	416 chroms_features[ feature.chrom ].append( feature )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	417
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	418 # Sort features by chrom, start position.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	419 chroms_features_sorted = []
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	420 for chrom_features in chroms_features.values():
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	421 chroms_features_sorted.append( chrom_features )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	422 chroms_features_sorted.sort( lambda a, b: cmp( a[0].chrom, b[0].chrom ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	423 for features in chroms_features_sorted:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	424 features.sort( lambda a, b: cmp( a.start, b.start ) )
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	425
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	426 # Yield comments first, then features.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	427 # FIXME: comments can appear anywhere in file, not just the beginning.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	428 # Ideally, then comments would be associated with features and output
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	429 # just before feature/line.
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	430 for comment in comments:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	431 yield comment
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	432
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	433 for chrom_features in chroms_features_sorted:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	434 for feature in chrom_features:
60925436ca5f planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/intersect commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689 devteam parents: diff changeset	435 yield feature

Mercurial > repos > devteam > intersect

annotate utils/gff_util.py @ 5:60925436ca5f draft default tip