# HG changeset patch # User devteam # Date 1396357973 14400 # Node ID 45fb880ab3bfd86e38c1d485af4abebbe968ccc2 Imported from capsule None diff -r 000000000000 -r 45fb880ab3bf cluster.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/cluster.xml Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,100 @@ + + the intervals of a dataset + + bx-python + galaxy-ops + + + gops_cluster.py $input1 $output -1 ${input1.metadata.chromCol},${input1.metadata.startCol},${input1.metadata.endCol},${input1.metadata.strandCol} -d $distance -m $minregions -o $returntype + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +.. class:: infomark + +**TIP:** If your dataset does not appear in the pulldown menu, it means that it is not in interval format. Use "edit attributes" to set chromosome, start, end, and strand columns. + +----- + +**Screencasts!** + +See Galaxy Interval Operation Screencasts_ (right click to open this link in another window). + +.. _Screencasts: http://wiki.g2.bx.psu.edu/Learn/Interval%20Operations + +----- + +**Syntax** + +- **Maximum distance** is greatest distance in base pairs allowed between intervals that will be considered "clustered". **Negative** values for distance are allowed, and are useful for clustering intervals that overlap. +- **Minimum intervals per cluster** allow a threshold to be set on the minimum number of intervals to be considered a cluster. Any area with less than this minimum will not be included in the output. +- **Merge clusters into single intervals** outputs intervals that span the entire cluster. +- **Find cluster intervals; preserve comments and order** filters out non-cluster intervals while maintaining the original ordering and comments in the file. +- **Find cluster intervals; output grouped by clusters** filters out non-cluster intervals, but outputs the cluster intervals so that they are grouped together. Comments and original ordering in the file are lost. + +----- + +**Examples** + +Find Clusters: + +.. image:: ${static_path}/operation_icons/gops_clusterFind.gif + +Merge Clusters: + +.. image:: ${static_path}/operation_icons/gops_clusterMerge.gif + + + diff -r 000000000000 -r 45fb880ab3bf gops_cluster.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/gops_cluster.py Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,129 @@ +#!/usr/bin/env python +""" +Cluster regions of intervals. + +usage: %prog in_file out_file + -1, --cols1=N,N,N,N: Columns for start, end, strand in file + -d, --distance=N: Maximum distance between clustered intervals + -v, --overlap=N: Minimum overlap require (negative distance) + -m, --minregions=N: Minimum regions per cluster + -o, --output=N: 1)merged 2)filtered 3)clustered 4) minimum 5) maximum +""" +import sys, traceback, fileinput +from warnings import warn +from bx.intervals import * +from bx.intervals.io import * +from bx.intervals.operations.find_clusters import * +from bx.cookbook import doc_optparse +from galaxy.tools.util.galaxyops import * + +assert sys.version_info[:2] >= ( 2, 4 ) + +def main(): + distance = 0 + minregions = 2 + output = 1 + upstream_pad = 0 + downstream_pad = 0 + + options, args = doc_optparse.parse( __doc__ ) + try: + chr_col_1, start_col_1, end_col_1, strand_col_1 = parse_cols_arg( options.cols1 ) + if options.distance: distance = int( options.distance ) + if options.overlap: distance = -1 * int( options.overlap ) + if options.output: output = int( options.output ) + if options.minregions: minregions = int( options.minregions ) + in_fname, out_fname = args + except: + doc_optparse.exception() + + g1 = NiceReaderWrapper( fileinput.FileInput( in_fname ), + chrom_col=chr_col_1, + start_col=start_col_1, + end_col=end_col_1, + strand_col=strand_col_1, + fix_strand=True ) + + # Get the cluster tree + try: + clusters, extra = find_clusters( g1, mincols=distance, minregions=minregions) + except ParseError, exc: + fail( "Invalid file format: %s" % str( exc ) ) + + f1 = open( in_fname, "r" ) + out_file = open( out_fname, "w" ) + + # If "merge" + if output == 1: + fields = ["." for x in range(max(g1.chrom_col, g1.start_col, g1.end_col)+1)] + for chrom, tree in clusters.items(): + for start, end, lines in tree.getregions(): + fields[g1.chrom_col] = chrom + fields[g1.start_col] = str(start) + fields[g1.end_col] = str(end) + out_file.write( "%s\n" % "\t".join( fields ) ) + + # If "filtered" we preserve order of file and comments, etc. + if output == 2: + linenums = dict() + for chrom, tree in clusters.items(): + for linenum in tree.getlines(): + linenums[linenum] = 0 + linenum = -1 + f1.seek(0) + for line in f1.readlines(): + linenum += 1 + if linenum in linenums or linenum in extra: + out_file.write( "%s\n" % line.rstrip( "\n\r" ) ) + + # If "clustered" we output original intervals, but near each other (i.e. clustered) + if output == 3: + linenums = list() + f1.seek(0) + fileLines = f1.readlines() + for chrom, tree in clusters.items(): + for linenum in tree.getlines(): + out_file.write( "%s\n" % fileLines[linenum].rstrip( "\n\r" ) ) + + # If "minimum" we output the smallest interval in each cluster + if output == 4 or output == 5: + linenums = list() + f1.seek(0) + fileLines = f1.readlines() + for chrom, tree in clusters.items(): + regions = tree.getregions() + for start, end, lines in tree.getregions(): + outsize = -1 + outinterval = None + for line in lines: + # three nested for loops? + # should only execute this code once per line + fileline = fileLines[line].rstrip("\n\r") + try: + cluster_interval = GenomicInterval( g1, fileline.split("\t"), + g1.chrom_col, + g1.start_col, + g1.end_col, + g1.strand_col, + g1.default_strand, + g1.fix_strand ) + except Exception, exc: + print >> sys.stderr, str( exc ) + f1.close() + sys.exit() + interval_size = cluster_interval.end - cluster_interval.start + if outsize == -1 or \ + ( outsize > interval_size and output == 4 ) or \ + ( outsize < interval_size and output == 5 ) : + outinterval = cluster_interval + outsize = interval_size + out_file.write( "%s\n" % outinterval ) + + f1.close() + out_file.close() + + if g1.skipped > 0: + print skipped( g1, filedesc="" ) + +if __name__ == "__main__": + main() diff -r 000000000000 -r 45fb880ab3bf operation_filter.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/operation_filter.py Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,99 @@ +# runs after the job (and after the default post-filter) +import os +from galaxy import eggs +from galaxy import jobs +from galaxy.tools.parameters import DataToolParameter + +from galaxy.jobs.handler import JOB_ERROR + +# Older py compatibility +try: + set() +except: + from sets import Set as set + +#def exec_before_process(app, inp_data, out_data, param_dict, tool=None): +# """Sets the name of the data""" +# dbkeys = sets.Set( [data.dbkey for data in inp_data.values() ] ) +# if len(dbkeys) != 1: +# raise Exception, '

Both Queries must be from the same genome build

' + +def validate_input( trans, error_map, param_values, page_param_map ): + dbkeys = set() + data_param_names = set() + data_params = 0 + for name, param in page_param_map.iteritems(): + if isinstance( param, DataToolParameter ): + # for each dataset parameter + if param_values.get(name, None) != None: + dbkeys.add( param_values[name].dbkey ) + data_params += 1 + # check meta data + try: + param = param_values[name] + if isinstance( param.datatype, trans.app.datatypes_registry.get_datatype_by_extension( 'gff' ).__class__ ): + # TODO: currently cannot validate GFF inputs b/c they are not derived from interval. + pass + else: # Validate interval datatype. + startCol = int( param.metadata.startCol ) + endCol = int( param.metadata.endCol ) + chromCol = int( param.metadata.chromCol ) + if param.metadata.strandCol is not None: + strandCol = int ( param.metadata.strandCol ) + else: + strandCol = 0 + except: + error_msg = "The attributes of this dataset are not properly set. " + \ + "Click the pencil icon in the history item to set the chrom, start, end and strand columns." + error_map[name] = error_msg + data_param_names.add( name ) + if len( dbkeys ) > 1: + for name in data_param_names: + error_map[name] = "All datasets must belong to same genomic build, " \ + "this dataset is linked to build '%s'" % param_values[name].dbkey + if data_params != len(data_param_names): + for name in data_param_names: + error_map[name] = "A dataset of the appropriate type is required" + +# Commented out by INS, 5/30/2007. What is the PURPOSE of this? +def exec_after_process(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + """Verify the output data after each run""" + items = out_data.items() + + for name, data in items: + try: + if stderr and len( stderr ) > 0: + raise Exception( stderr ) + + except Exception, exc: + data.blurb = JOB_ERROR + data.state = JOB_ERROR + +## def exec_after_process(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): +## pass + + +def exec_after_merge(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + items = out_data.items() + for name, data in items: + if param_dict['returntype'] == True: + data.metadata.chromCol = 1 + data.metadata.startCol = 2 + data.metadata.endCol = 3 + # merge always clobbers strand + data.metadata.strandCol = None + + +def exec_after_cluster(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + if param_dict["returntype"] == '1': + items = out_data.items() + for name, data in items: + data.metadata.strandCol = None diff -r 000000000000 -r 45fb880ab3bf test-data/5.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/5.bed Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,134 @@ +chr7 115444712 115444739 CCDS5763.1_cds_0_0_chr7_115444713_f 0 + +chr7 115468538 115468624 CCDS5763.1_cds_1_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5763.1_cds_2_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5763.1_cds_3_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5763.1_cds_4_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5763.1_cds_5_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5763.1_cds_6_0_chr7_115491299_f 0 + +chr7 115468538 115468624 CCDS5764.1_cds_0_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5764.1_cds_1_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5764.1_cds_2_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5764.1_cds_3_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5764.1_cds_4_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5764.1_cds_5_0_chr7_115491299_f 0 + +chr7 115733786 115733936 CCDS5766.1_cds_0_0_chr7_115733787_f 0 + +chr7 115734264 115734452 CCDS5766.1_cds_1_0_chr7_115734265_f 0 + +chr7 115739975 115740126 CCDS5766.1_cds_2_0_chr7_115739976_f 0 + +chr7 115733786 115733936 CCDS5765.1_cds_0_0_chr7_115733787_f 0 + +chr7 115739975 115740164 CCDS5765.1_cds_1_0_chr7_115739976_f 0 + +chr7 115759067 115759097 CCDS5767.1_cds_0_0_chr7_115759068_f 0 + +chr7 115760529 115760694 CCDS5767.1_cds_1_0_chr7_115760530_f 0 + +chr7 115792950 115793292 CCDS5767.1_cds_2_0_chr7_115792951_f 0 + +chr7 116096616 116096655 CCDS5768.1_cds_0_0_chr7_116096617_f 0 + +chr7 116122131 116122195 CCDS5768.1_cds_1_0_chr7_116122132_f 0 + +chr7 116126998 116127050 CCDS5768.1_cds_2_0_chr7_116126999_f 0 + +chr7 116132776 116132840 CCDS5768.1_cds_3_0_chr7_116132777_f 0 + +chr7 116138181 116138388 CCDS5768.1_cds_4_0_chr7_116138182_f 0 + +chr7 116140267 116140347 CCDS5768.1_cds_5_0_chr7_116140268_f 0 + +chr7 116144237 116144316 CCDS5768.1_cds_6_0_chr7_116144238_f 0 + +chr7 116146073 116146145 CCDS5768.1_cds_7_0_chr7_116146074_f 0 + +chr7 116150064 116150127 CCDS5768.1_cds_8_0_chr7_116150065_f 0 + +chr7 116151731 116151872 CCDS5768.1_cds_9_0_chr7_116151732_f 0 + +chr7 116187545 116187696 CCDS5770.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5770.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5770.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5770.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5770.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5770.1_cds_5_0_chr7_116365890_f 0 + +chr7 116368128 116368197 CCDS5770.1_cds_6_0_chr7_116368129_f 0 + +chr7 116370085 116370240 CCDS5770.1_cds_7_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5770.1_cds_8_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5770.1_cds_9_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5770.1_cds_10_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5770.1_cds_11_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5770.1_cds_12_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5770.1_cds_13_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5770.1_cds_14_0_chr7_116455928_f 0 + +chr7 116456865 116456985 CCDS5770.1_cds_15_0_chr7_116456866_f 0 + +chr7 116187545 116187696 CCDS5769.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5769.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5769.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5769.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5769.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5769.1_cds_5_0_chr7_116365890_f 0 + +chr7 116370085 116370240 CCDS5769.1_cds_6_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5769.1_cds_7_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5769.1_cds_8_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5769.1_cds_9_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5769.1_cds_10_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5769.1_cds_11_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5769.1_cds_12_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5769.1_cds_13_0_chr7_116455928_f 0 + +chr7 116463766 116463862 CCDS5769.1_cds_14_0_chr7_116463767_f 0 + +chr7 116512159 116512389 CCDS5771.1_cds_0_0_chr7_116512160_r 0 - +chr7 116531616 116531881 CCDS5771.1_cds_1_0_chr7_116531617_r 0 - +chr7 116549075 116549353 CCDS5771.1_cds_2_0_chr7_116549076_r 0 - +chr7 116554571 116554798 CCDS5771.1_cds_3_0_chr7_116554572_r 0 - +chr7 116556911 116556994 CCDS5771.1_cds_4_0_chr7_116556912_r 0 - +chr7 116597600 116597753 CCDS5772.1_cds_0_0_chr7_116597601_r 0 - +chr7 116601356 116601470 CCDS5772.1_cds_1_0_chr7_116601357_r 0 - +chr7 116602616 116602722 CCDS5772.1_cds_2_0_chr7_116602617_r 0 - +chr7 116613942 116614052 CCDS5772.1_cds_3_0_chr7_116613943_r 0 - +chr7 116615015 116615072 CCDS5772.1_cds_4_0_chr7_116615016_r 0 - +chr7 116616073 116616149 CCDS5772.1_cds_5_0_chr7_116616074_r 0 - +chr7 116616990 116617115 CCDS5772.1_cds_6_0_chr7_116616991_r 0 - +chr7 116618730 116618865 CCDS5772.1_cds_7_0_chr7_116618731_r 0 - +chr7 116619702 116619814 CCDS5772.1_cds_8_0_chr7_116619703_r 0 - +chr7 116654167 116654279 CCDS5772.1_cds_9_0_chr7_116654168_r 0 - +chr7 116656241 116656364 CCDS5772.1_cds_10_0_chr7_116656242_r 0 - +chr7 116660840 116660940 CCDS5772.1_cds_11_0_chr7_116660841_r 0 - +chr7 116661360 116661465 CCDS5772.1_cds_12_0_chr7_116661361_r 0 - +chr7 116714099 116714152 CCDS5773.1_cds_0_0_chr7_116714100_f 0 + +chr7 116738257 116738368 CCDS5773.1_cds_1_0_chr7_116738258_f 0 + +chr7 116743038 116743147 CCDS5773.1_cds_2_0_chr7_116743039_f 0 + +chr7 116764903 116765119 CCDS5773.1_cds_3_0_chr7_116764904_f 0 + +chr7 116768280 116768370 CCDS5773.1_cds_4_0_chr7_116768281_f 0 + +chr7 116769252 116769416 CCDS5773.1_cds_5_0_chr7_116769253_f 0 + +chr7 116770552 116770678 CCDS5773.1_cds_6_0_chr7_116770553_f 0 + +chr7 116774104 116774351 CCDS5773.1_cds_7_0_chr7_116774105_f 0 + +chr7 116776020 116776113 CCDS5773.1_cds_8_0_chr7_116776021_f 0 + +chr7 116782645 116782828 CCDS5773.1_cds_9_0_chr7_116782646_f 0 + +chr7 116793468 116793660 CCDS5773.1_cds_10_0_chr7_116793469_f 0 + +chr7 116821743 116821838 CCDS5773.1_cds_11_0_chr7_116821744_f 0 + +chr7 116824357 116824444 CCDS5773.1_cds_12_0_chr7_116824358_f 0 + +chr7 116825938 116826662 CCDS5773.1_cds_13_0_chr7_116825939_f 0 + +chr7 116828934 116829063 CCDS5773.1_cds_14_0_chr7_116828935_f 0 + +chr7 116836830 116836868 CCDS5773.1_cds_15_0_chr7_116836831_f 0 + +chr7 116837536 116837787 CCDS5773.1_cds_16_0_chr7_116837537_f 0 + +chr7 116840678 116840758 CCDS5773.1_cds_17_0_chr7_116840679_f 0 + +chr7 116844523 116844674 CCDS5773.1_cds_18_0_chr7_116844524_f 0 + +chr7 116845585 116845813 CCDS5773.1_cds_19_0_chr7_116845586_f 0 + +chr7 116848617 116848718 CCDS5773.1_cds_20_0_chr7_116848618_f 0 + +chr7 116861526 116861775 CCDS5773.1_cds_21_0_chr7_116861527_f 0 + +chr7 116876442 116876598 CCDS5773.1_cds_22_0_chr7_116876443_f 0 + +chr7 116886846 116886936 CCDS5773.1_cds_23_0_chr7_116886847_f 0 + +chr7 116898692 116898865 CCDS5773.1_cds_24_0_chr7_116898693_f 0 + +chr7 116899463 116899569 CCDS5773.1_cds_25_0_chr7_116899464_f 0 + +chr7 116900912 116901113 CCDS5773.1_cds_26_0_chr7_116900913_f 0 + +chr7 116945541 116945787 CCDS5774.1_cds_0_0_chr7_116945542_r 0 - +chr7 116952022 116952124 CCDS5774.1_cds_1_0_chr7_116952023_r 0 - +chr7 116953508 116953641 CCDS5774.1_cds_2_0_chr7_116953509_r 0 - +chr7 116955071 116955135 CCDS5774.1_cds_3_0_chr7_116955072_r 0 - +chr7 116958551 116958737 CCDS5774.1_cds_4_0_chr7_116958552_r 0 - +chr7 116959056 116959262 CCDS5774.1_cds_5_0_chr7_116959057_r 0 - +chr7 116962093 116962272 CCDS5774.1_cds_6_0_chr7_116962094_r 0 - +chr7 116968917 116969105 CCDS5774.1_cds_7_0_chr7_116968918_r 0 - +chr7 116969273 116969426 CCDS5774.1_cds_8_0_chr7_116969274_r 0 - +chr7 116979835 116979935 CCDS5774.1_cds_9_0_chr7_116979836_r 0 - +chr7 116980017 116980104 CCDS5774.1_cds_10_0_chr7_116980018_r 0 - +chr7 116990559 116990639 CCDS5774.1_cds_11_0_chr7_116990560_r 0 - +chr7 116991879 116991975 CCDS5774.1_cds_12_0_chr7_116991880_r 0 - +chr7 116994439 116994715 CCDS5774.1_cds_13_0_chr7_116994440_r 0 - +chr7 117001063 117001181 CCDS5774.1_cds_14_0_chr7_117001064_r 0 - +chr7 117011515 117011770 CCDS5774.1_cds_15_0_chr7_117011516_r 0 - +chr7 117014445 117014596 CCDS5774.1_cds_16_0_chr7_117014446_r 0 - +chr7 117016866 117016966 CCDS5774.1_cds_17_0_chr7_117016867_r 0 - +chr7 117018255 117018459 CCDS5774.1_cds_18_0_chr7_117018256_r 0 - +chr7 117025132 117026786 CCDS5774.1_cds_19_0_chr7_117025133_r 0 - +chr7 117044769 117044994 CCDS5774.1_cds_20_0_chr7_117044770_r 0 - +chr7 117095213 117095321 CCDS5774.1_cds_21_0_chr7_117095214_r 0 - +chr7 117107339 117107420 CCDS5774.1_cds_22_0_chr7_117107340_r 0 - +chr5 131424298 131424460 CCDS4149.1_cds_0_0_chr5_131424299_f 0 + +chr5 131424558 131424600 CCDS4149.1_cds_1_0_chr5_131424559_f 0 + +chr5 131425903 131425993 CCDS4149.1_cds_2_0_chr5_131425904_f 0 + +chr5 131426117 131426159 CCDS4149.1_cds_3_0_chr5_131426118_f 0 + diff -r 000000000000 -r 45fb880ab3bf test-data/gops-cluster-1.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/gops-cluster-1.bed Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,22 @@ +chr7 115468538 115468624 +chr7 115483024 115483277 +chr7 115484165 115484501 +chr7 115485764 115485980 +chr7 115486322 115486481 +chr7 115491298 115491487 +chr7 115733786 115733936 +chr7 115739975 115740164 +chr7 116187545 116187696 +chr7 116333766 116333849 +chr7 116353565 116353725 +chr7 116363797 116363852 +chr7 116364495 116364611 +chr7 116365889 116365965 +chr7 116370085 116370240 +chr7 116372439 116372537 +chr7 116404866 116404981 +chr7 116423325 116423398 +chr7 116424838 116424941 +chr7 116443791 116443942 +chr7 116453088 116453181 +chr7 116455927 116456067 diff -r 000000000000 -r 45fb880ab3bf test-data/gops-cluster-2.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/gops-cluster-2.bed Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,44 @@ +chr7 115468538 115468624 CCDS5763.1_cds_1_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5763.1_cds_2_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5763.1_cds_3_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5763.1_cds_4_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5763.1_cds_5_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5763.1_cds_6_0_chr7_115491299_f 0 + +chr7 115468538 115468624 CCDS5764.1_cds_0_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5764.1_cds_1_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5764.1_cds_2_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5764.1_cds_3_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5764.1_cds_4_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5764.1_cds_5_0_chr7_115491299_f 0 + +chr7 115733786 115733936 CCDS5766.1_cds_0_0_chr7_115733787_f 0 + +chr7 115739975 115740126 CCDS5766.1_cds_2_0_chr7_115739976_f 0 + +chr7 115733786 115733936 CCDS5765.1_cds_0_0_chr7_115733787_f 0 + +chr7 115739975 115740164 CCDS5765.1_cds_1_0_chr7_115739976_f 0 + +chr7 116187545 116187696 CCDS5770.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5770.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5770.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5770.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5770.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5770.1_cds_5_0_chr7_116365890_f 0 + +chr7 116370085 116370240 CCDS5770.1_cds_7_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5770.1_cds_8_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5770.1_cds_9_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5770.1_cds_10_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5770.1_cds_11_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5770.1_cds_12_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5770.1_cds_13_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5770.1_cds_14_0_chr7_116455928_f 0 + +chr7 116187545 116187696 CCDS5769.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5769.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5769.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5769.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5769.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5769.1_cds_5_0_chr7_116365890_f 0 + +chr7 116370085 116370240 CCDS5769.1_cds_6_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5769.1_cds_7_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5769.1_cds_8_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5769.1_cds_9_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5769.1_cds_10_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5769.1_cds_11_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5769.1_cds_12_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5769.1_cds_13_0_chr7_116455928_f 0 + diff -r 000000000000 -r 45fb880ab3bf test-data/gops-cluster-3.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/gops-cluster-3.bed Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,44 @@ +chr7 115468538 115468624 CCDS5763.1_cds_1_0_chr7_115468539_f 0 + +chr7 115468538 115468624 CCDS5764.1_cds_0_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5763.1_cds_2_0_chr7_115483025_f 0 + +chr7 115483024 115483277 CCDS5764.1_cds_1_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5763.1_cds_3_0_chr7_115484166_f 0 + +chr7 115484165 115484501 CCDS5764.1_cds_2_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5763.1_cds_4_0_chr7_115485765_f 0 + +chr7 115485764 115485980 CCDS5764.1_cds_3_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5763.1_cds_5_0_chr7_115486323_f 0 + +chr7 115486322 115486481 CCDS5764.1_cds_4_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5763.1_cds_6_0_chr7_115491299_f 0 + +chr7 115491298 115491487 CCDS5764.1_cds_5_0_chr7_115491299_f 0 + +chr7 115733786 115733936 CCDS5766.1_cds_0_0_chr7_115733787_f 0 + +chr7 115733786 115733936 CCDS5765.1_cds_0_0_chr7_115733787_f 0 + +chr7 115739975 115740126 CCDS5766.1_cds_2_0_chr7_115739976_f 0 + +chr7 115739975 115740164 CCDS5765.1_cds_1_0_chr7_115739976_f 0 + +chr7 116187545 116187696 CCDS5770.1_cds_0_0_chr7_116187546_f 0 + +chr7 116187545 116187696 CCDS5769.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5770.1_cds_1_0_chr7_116333767_f 0 + +chr7 116333766 116333849 CCDS5769.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5770.1_cds_2_0_chr7_116353566_f 0 + +chr7 116353565 116353725 CCDS5769.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5770.1_cds_3_0_chr7_116363798_f 0 + +chr7 116363797 116363852 CCDS5769.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5770.1_cds_4_0_chr7_116364496_f 0 + +chr7 116364495 116364611 CCDS5769.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5770.1_cds_5_0_chr7_116365890_f 0 + +chr7 116365889 116365965 CCDS5769.1_cds_5_0_chr7_116365890_f 0 + +chr7 116370085 116370240 CCDS5770.1_cds_7_0_chr7_116370086_f 0 + +chr7 116370085 116370240 CCDS5769.1_cds_6_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5770.1_cds_8_0_chr7_116372440_f 0 + +chr7 116372439 116372537 CCDS5769.1_cds_7_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5770.1_cds_9_0_chr7_116404867_f 0 + +chr7 116404866 116404981 CCDS5769.1_cds_8_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5770.1_cds_10_0_chr7_116423326_f 0 + +chr7 116423325 116423398 CCDS5769.1_cds_9_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5770.1_cds_11_0_chr7_116424839_f 0 + +chr7 116424838 116424941 CCDS5769.1_cds_10_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5770.1_cds_12_0_chr7_116443792_f 0 + +chr7 116443791 116443942 CCDS5769.1_cds_11_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5770.1_cds_13_0_chr7_116453089_f 0 + +chr7 116453088 116453181 CCDS5769.1_cds_12_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5770.1_cds_14_0_chr7_116455928_f 0 + +chr7 116455927 116456067 CCDS5769.1_cds_13_0_chr7_116455928_f 0 + diff -r 000000000000 -r 45fb880ab3bf test-data/gops_cluster_bigint.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/gops_cluster_bigint.bed Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,135 @@ +chr7 115444712 115444739 CCDS5763.1_cds_0_0_chr7_115444713_f 0 + +chr7 115468538 115468624 CCDS5763.1_cds_1_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5763.1_cds_2_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5763.1_cds_3_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5763.1_cds_4_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5763.1_cds_5_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5763.1_cds_6_0_chr7_115491299_f 0 + +chr7 115468538 115468624 CCDS5764.1_cds_0_0_chr7_115468539_f 0 + +chr7 115483024 115483277 CCDS5764.1_cds_1_0_chr7_115483025_f 0 + +chr7 115484165 115484501 CCDS5764.1_cds_2_0_chr7_115484166_f 0 + +chr7 115485764 115485980 CCDS5764.1_cds_3_0_chr7_115485765_f 0 + +chr7 115486322 115486481 CCDS5764.1_cds_4_0_chr7_115486323_f 0 + +chr7 115491298 115491487 CCDS5764.1_cds_5_0_chr7_115491299_f 0 + +chr7 115733786 115733936 CCDS5766.1_cds_0_0_chr7_115733787_f 0 + +chr7 115734264 115734452 CCDS5766.1_cds_1_0_chr7_115734265_f 0 + +chr7 115739975 115740126 CCDS5766.1_cds_2_0_chr7_115739976_f 0 + +chr7 115733786 115733936 CCDS5765.1_cds_0_0_chr7_115733787_f 0 + +chr7 115739975 115740164 CCDS5765.1_cds_1_0_chr7_115739976_f 0 + +chr7 115759067 115759097 CCDS5767.1_cds_0_0_chr7_115759068_f 0 + +chr7 115760529 115760694 CCDS5767.1_cds_1_0_chr7_115760530_f 0 + +chr7 115792950 115793292 CCDS5767.1_cds_2_0_chr7_115792951_f 0 + +chr7 116096616 116096655 CCDS5768.1_cds_0_0_chr7_116096617_f 0 + +chr7 116122131 116122195 CCDS5768.1_cds_1_0_chr7_116122132_f 0 + +chr7 116126998 116127050 CCDS5768.1_cds_2_0_chr7_116126999_f 0 + +chr7 116132776 116132840 CCDS5768.1_cds_3_0_chr7_116132777_f 0 + +chr7 116138181 116138388 CCDS5768.1_cds_4_0_chr7_116138182_f 0 + +chr7 116140267 116140347 CCDS5768.1_cds_5_0_chr7_116140268_f 0 + +chr7 116144237 116144316 CCDS5768.1_cds_6_0_chr7_116144238_f 0 + +chr7 116146073 116146145 CCDS5768.1_cds_7_0_chr7_116146074_f 0 + +chr7 116150064 116150127 CCDS5768.1_cds_8_0_chr7_116150065_f 0 + +chr7 116151731 116151872 CCDS5768.1_cds_9_0_chr7_116151732_f 0 + +chr7 116187545 116187696 CCDS5770.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5770.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5770.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5770.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5770.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5770.1_cds_5_0_chr7_116365890_f 0 + +chr7 116368128 116368197 CCDS5770.1_cds_6_0_chr7_116368129_f 0 + +chr7 116370085 116370240 CCDS5770.1_cds_7_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5770.1_cds_8_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5770.1_cds_9_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5770.1_cds_10_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5770.1_cds_11_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5770.1_cds_12_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5770.1_cds_13_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5770.1_cds_14_0_chr7_116455928_f 0 + +chr7 116456865 116456985 CCDS5770.1_cds_15_0_chr7_116456866_f 0 + +chr7 116187545 116187696 CCDS5769.1_cds_0_0_chr7_116187546_f 0 + +chr7 116333766 116333849 CCDS5769.1_cds_1_0_chr7_116333767_f 0 + +chr7 116353565 116353725 CCDS5769.1_cds_2_0_chr7_116353566_f 0 + +chr7 116363797 116363852 CCDS5769.1_cds_3_0_chr7_116363798_f 0 + +chr7 116364495 116364611 CCDS5769.1_cds_4_0_chr7_116364496_f 0 + +chr7 116365889 116365965 CCDS5769.1_cds_5_0_chr7_116365890_f 0 + +chr7 116370085 116370240 CCDS5769.1_cds_6_0_chr7_116370086_f 0 + +chr7 116372439 116372537 CCDS5769.1_cds_7_0_chr7_116372440_f 0 + +chr7 116404866 116404981 CCDS5769.1_cds_8_0_chr7_116404867_f 0 + +chr7 116423325 116423398 CCDS5769.1_cds_9_0_chr7_116423326_f 0 + +chr7 116424838 116424941 CCDS5769.1_cds_10_0_chr7_116424839_f 0 + +chr7 116443791 116443942 CCDS5769.1_cds_11_0_chr7_116443792_f 0 + +chr7 116453088 116453181 CCDS5769.1_cds_12_0_chr7_116453089_f 0 + +chr7 116455927 116456067 CCDS5769.1_cds_13_0_chr7_116455928_f 0 + +chr7 116463766 116463862 CCDS5769.1_cds_14_0_chr7_116463767_f 0 + +chr7 116512159 116512389 CCDS5771.1_cds_0_0_chr7_116512160_r 0 - +chr7 116531616 116531881 CCDS5771.1_cds_1_0_chr7_116531617_r 0 - +chr7 116549075 116549353 CCDS5771.1_cds_2_0_chr7_116549076_r 0 - +chr7 116554571 116554798 CCDS5771.1_cds_3_0_chr7_116554572_r 0 - +chr7 116556911 116556994 CCDS5771.1_cds_4_0_chr7_116556912_r 0 - +chr7 116597600 116597753 CCDS5772.1_cds_0_0_chr7_116597601_r 0 - +chr7 116601356 116601470 CCDS5772.1_cds_1_0_chr7_116601357_r 0 - +chr7 116602616 116602722 CCDS5772.1_cds_2_0_chr7_116602617_r 0 - +chr7 116613942 116614052 CCDS5772.1_cds_3_0_chr7_116613943_r 0 - +chr7 116615015 116615072 CCDS5772.1_cds_4_0_chr7_116615016_r 0 - +chr7 116616073 116616149 CCDS5772.1_cds_5_0_chr7_116616074_r 0 - +chr7 116616990 116617115 CCDS5772.1_cds_6_0_chr7_116616991_r 0 - +chr7 116618730 116618865 CCDS5772.1_cds_7_0_chr7_116618731_r 0 - +chr7 116619702 116619814 CCDS5772.1_cds_8_0_chr7_116619703_r 0 - +chr7 116654167 116654279 CCDS5772.1_cds_9_0_chr7_116654168_r 0 - +chr7 116656241 116656364 CCDS5772.1_cds_10_0_chr7_116656242_r 0 - +chr7 116660840 116660940 CCDS5772.1_cds_11_0_chr7_116660841_r 0 - +chr7 116661360 116661465 CCDS5772.1_cds_12_0_chr7_116661361_r 0 - +chr7 116714099 116714152 CCDS5773.1_cds_0_0_chr7_116714100_f 0 + +chr7 116738257 116738368 CCDS5773.1_cds_1_0_chr7_116738258_f 0 + +chr7 116743038 116743147 CCDS5773.1_cds_2_0_chr7_116743039_f 0 + +chr7 116764903 116765119 CCDS5773.1_cds_3_0_chr7_116764904_f 0 + +chr7 116768280 116768370 CCDS5773.1_cds_4_0_chr7_116768281_f 0 + +chr7 116769252 116769416 CCDS5773.1_cds_5_0_chr7_116769253_f 0 + +chr7 116770552 116770678 CCDS5773.1_cds_6_0_chr7_116770553_f 0 + +chr7 116774104 116774351 CCDS5773.1_cds_7_0_chr7_116774105_f 0 + +chr7 116776020 116776113 CCDS5773.1_cds_8_0_chr7_116776021_f 0 + +chr7 116782645 116782828 CCDS5773.1_cds_9_0_chr7_116782646_f 0 + +chr7 116793468 116793660 CCDS5773.1_cds_10_0_chr7_116793469_f 0 + +chr7 116821743 116821838 CCDS5773.1_cds_11_0_chr7_116821744_f 0 + +chr7 116824357 116824444 CCDS5773.1_cds_12_0_chr7_116824358_f 0 + +chr7 116825938 116826662 CCDS5773.1_cds_13_0_chr7_116825939_f 0 + +chr7 116828934 116829063 CCDS5773.1_cds_14_0_chr7_116828935_f 0 + +chr7 116836830 116836868 CCDS5773.1_cds_15_0_chr7_116836831_f 0 + +chr7 116837536 116837787 CCDS5773.1_cds_16_0_chr7_116837537_f 0 + +chr7 116840678 116840758 CCDS5773.1_cds_17_0_chr7_116840679_f 0 + +chr7 116844523 116844674 CCDS5773.1_cds_18_0_chr7_116844524_f 0 + +chr7 116845585 116845813 CCDS5773.1_cds_19_0_chr7_116845586_f 0 + +chr7 116848617 116848718 CCDS5773.1_cds_20_0_chr7_116848618_f 0 + +chr7 116861526 116861775 CCDS5773.1_cds_21_0_chr7_116861527_f 0 + +chr7 116876442 116876598 CCDS5773.1_cds_22_0_chr7_116876443_f 0 + +chr7 116886846 116886936 CCDS5773.1_cds_23_0_chr7_116886847_f 0 + +chr7 116898692 116898865 CCDS5773.1_cds_24_0_chr7_116898693_f 0 + +chr7 116899463 116899569 CCDS5773.1_cds_25_0_chr7_116899464_f 0 + +chr7 116900912 116901113 CCDS5773.1_cds_26_0_chr7_116900913_f 0 + +chr7 116945541 116945787 CCDS5774.1_cds_0_0_chr7_116945542_r 0 - +chr7 116952022 116952124 CCDS5774.1_cds_1_0_chr7_116952023_r 0 - +chr7 116953508 116953641 CCDS5774.1_cds_2_0_chr7_116953509_r 0 - +chr7 116955071 116955135 CCDS5774.1_cds_3_0_chr7_116955072_r 0 - +chr7 116958551 116958737 CCDS5774.1_cds_4_0_chr7_116958552_r 0 - +chr7 116959056 116959262 CCDS5774.1_cds_5_0_chr7_116959057_r 0 - +chr7 116962093 116962272 CCDS5774.1_cds_6_0_chr7_116962094_r 0 - +chr7 116968917 116969105 CCDS5774.1_cds_7_0_chr7_116968918_r 0 - +chr7 116969273 116969426 CCDS5774.1_cds_8_0_chr7_116969274_r 0 - +chr7 116979835 116979935 CCDS5774.1_cds_9_0_chr7_116979836_r 0 - +chr7 116980017 116980104 CCDS5774.1_cds_10_0_chr7_116980018_r 0 - +chr7 116990559 116990639 CCDS5774.1_cds_11_0_chr7_116990560_r 0 - +chr7 116991879 116991975 CCDS5774.1_cds_12_0_chr7_116991880_r 0 - +chr7 116994439 116994715 CCDS5774.1_cds_13_0_chr7_116994440_r 0 - +chr7 117001063 117001181 CCDS5774.1_cds_14_0_chr7_117001064_r 0 - +chr7 117011515 117011770 CCDS5774.1_cds_15_0_chr7_117011516_r 0 - +chr7 117014445 117014596 CCDS5774.1_cds_16_0_chr7_117014446_r 0 - +chr7 117016866 117016966 CCDS5774.1_cds_17_0_chr7_117016867_r 0 - +chr7 117018255 117018459 CCDS5774.1_cds_18_0_chr7_117018256_r 0 - +chr7 117025132 117026786 CCDS5774.1_cds_19_0_chr7_117025133_r 0 - +chr7 117044769 117044994 CCDS5774.1_cds_20_0_chr7_117044770_r 0 - +chr7 117095213 117095321 CCDS5774.1_cds_21_0_chr7_117095214_r 0 - +chr7 117107339 117107420 CCDS5774.1_cds_22_0_chr7_117107340_r 0 - +chr5 131424298 131424460 CCDS4149.1_cds_0_0_chr5_131424299_f 0 + +chr5 131424558 131424600 CCDS4149.1_cds_1_0_chr5_131424559_f 0 + +chr5 131425903 131425993 CCDS4149.1_cds_2_0_chr5_131425904_f 0 + +chr5 131426117 131426159 CCDS4149.1_cds_3_0_chr5_131426118_f 0 + +chrM 4294967295 4294967322 CCDS4149.1_cds_4_0_chr5_4294967296_f 0 + \ No newline at end of file diff -r 000000000000 -r 45fb880ab3bf tool_dependencies.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tool_dependencies.xml Tue Apr 01 09:12:53 2014 -0400 @@ -0,0 +1,9 @@ + + + + + + + + +