comparison bowtie_wrapper.py @ 4:bdd9fcdb398e draft

planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tools/bowtie_wrappers commit 397bc85634156862fdaa3d74b8f1e53752876619

3:9326d6e6882a

    -x, --indexSettings=x: Whether or not indexing options are to be set
    -H, --suppressHeader=H: Suppress header
    --do_not_build_index: Flag to specify that provided file is already indexed and to just use 'as is'
"""

import optparse, os, shutil, subprocess, sys, tempfile

#Allow more than Sanger encoded variants
DEFAULT_ASCII_ENCODING = '--phred33-quals'
GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG = { 'fastqsanger':'--phred33-quals', 'fastqillumina':'--phred64-quals', 'fastqsolexa':'--solexa-quals' }
#FIXME: Integer quality scores are supported only when the '--integer-quals' argument is specified to bowtie; this is not currently able to be set in the tool/wrapper/config

def stop_err( msg ):
    sys.stderr.write( '%s\n' % msg )
    sys.exit()

def __main__():
    #Parse Command Line
    parser = optparse.OptionParser()
    parser.add_option( '-t', '--threads', dest='threads', help='The number of threads to run' )
    parser.add_option( '-o', '--output', dest='output', help='The output file' )
    parser.add_option( '', '--output_unmapped_reads', dest='output_unmapped_reads', help='File name for unmapped reads (single-end)' )
    parser.add_option( '', '--output_unmapped_reads_l', dest='output_unmapped_reads_l', help='File name for unmapped reads (left, paired-end)' )

    else:
        colorspace = ''
    # index if necessary
    if options.genomeSource == 'history' and not options.do_not_build_index:
        # set up commands
        if options.index_settings =='indexPreSet':
            indexing_cmds = '%s' % colorspace
        else:
            try:
                if options.iautoB and options.iautoB == 'set':
                    iautoB = '--noauto'

                    iseed = ''
                indexing_cmds = '%s %s %s %s %s %s %s --offrate %s %s %s %s %s %s' % \
                                ( iautoB, ipacked, ibmax, ibmaxdivn, idcv, inodc,
                                  inoref, options.ioffrate, iftab, intoa, iendian,
                                  iseed, colorspace )
            except ValueError, e:
                # clean up temp dir
                if os.path.exists( tmp_index_dir ):
                    shutil.rmtree( tmp_index_dir )
                stop_err( "Something is wrong with the indexing parameters and the indexing and alignment could not be run. Make sure you don't have any non-numeric values where they should be numeric.\n" + str( e ) )
        ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )

        ref_file.close()
        os.symlink( options.ref, ref_file_name )
        cmd1 = 'bowtie-build %s -f %s %s' % ( indexing_cmds, ref_file_name, ref_file_name )
        try:
            tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
            tmp_stderr = open( tmp, 'wb' )
            proc = subprocess.Popen( args=cmd1, shell=True, cwd=tmp_index_dir, stderr=tmp_stderr.fileno() )
            returncode = proc.wait()
            tmp_stderr.close()
            # get stderr, allowing for case where it's very large
            tmp_stderr = open( tmp, 'rb' )
            stderr = ''
            buffsize = 1048576
            try:
                while True:
                    stderr += tmp_stderr.read( buffsize )
                    if not stderr or len( stderr ) % buffsize != 0:
                        break
            except OverflowError:
                pass
            tmp_stderr.close()
            if returncode != 0:
                raise Exception, stderr
        except Exception, e:
            # clean up temp dir
            if os.path.exists( tmp_index_dir ):
                shutil.rmtree( tmp_index_dir )
            stop_err( 'Error indexing reference sequence\n' + str( e ) )
        stdout += 'File indexed. '

    else:
        mateOrient = ''
    quality_score_encoding = GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG.get( options.galaxy_input_format, DEFAULT_ASCII_ENCODING )
    if options.params == 'preSet':
        aligning_cmds = '-q %s %s -p %s -S %s %s %s ' % \
                ( maxInsert, mateOrient, options.threads, suppressHeader, colorspace, quality_score_encoding )
    else:
        try:
            if options.skip and int( options.skip ) > 0:
                skip = '-s %s' % options.skip
            else:

                trimH = ''
            if options.trimL and int( options.trimL ) > 0:
                trimL = '-3 %s' % options.trimL
            else:
                trimL = ''
            if options.maxMismatches and (options.maxMismatches == '0' or options.maxMismatches == '1' \
                        or options.maxMismatches == '2' or options.maxMismatches == '3'):
                maxMismatches = '-v %s' % options.maxMismatches
            else:
                maxMismatches = ''
            if options.mismatchSeed and (options.mismatchSeed == '0' or options.mismatchSeed == '1' \
                        or options.mismatchSeed == '2' or options.mismatchSeed == '3'):
                mismatchSeed = '-n %s' % options.mismatchSeed
            else:
                mismatchSeed = ''
            if options.mismatchQual and int( options.mismatchQual ) >= 1:
                mismatchQual = '-e %s' % options.mismatchQual

                              maxAlignAttempt, forwardAlign, reverseAlign, maxBacktracks,
                              tryHard, valAlign, allValAligns, suppressAlign, best,
                              strata, offrate, seed, snpphred, snpfrac, keepends,
                              output_unmapped_reads, output_suppressed_reads,
                              quality_score_encoding )
        except ValueError, e:
            # clean up temp dir
            if os.path.exists( tmp_index_dir ):
                shutil.rmtree( tmp_index_dir )
            stop_err( 'Something is wrong with the alignment parameters and the alignment could not be run\n' + str( e ) )
    try:

                cmd2 = 'bowtie %s %s -1 %s -2 %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.input2, options.output )
            else:
                cmd2 = 'bowtie %s %s %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.output )
            # align
            tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
            tmp_stderr = open( tmp, 'wb' )
            proc = subprocess.Popen( args=cmd2, shell=True, cwd=tmp_index_dir, stdout=sys.stdout, stderr=tmp_stderr.fileno() )
            returncode = proc.wait()
            tmp_stderr.close()
            # get stderr, allowing for case where it's very large
            tmp_stderr = open( tmp, 'rb' )
            stderr = ''
            buffsize = 1048576
            try:
                while True:
                    stderr += tmp_stderr.read( buffsize )
                    if not stderr or len( stderr ) % buffsize != 0:
                        break
            except OverflowError:
                pass
            tmp_stderr.close()
            if returncode != 0:
                raise Exception, stderr
            elif options.output_mapping_stats is not None:
                # Write stderr (containing the mapping statistics) to a named file
                with open(options.output_mapping_stats, 'w') as mapping_stats:
                    mapping_stats.write( stderr )
            # get suppressed and unmapped reads output files in place if appropriate
            if options.paired == 'paired' and tmp_suppressed_file_name and \
                               options.output_suppressed_reads_l and options.output_suppressed_reads_r:
                try:
                    left = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
                    right = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
                    shutil.move( left, options.output_suppressed_reads_l )
                    shutil.move( right, options.output_suppressed_reads_r )
                except Exception, e:
                    sys.stdout.write( 'Error producing the suppressed output file.\n' )
            if options.paired == 'paired' and tmp_unmapped_file_name and \
                               options.output_unmapped_reads_l and options.output_unmapped_reads_r:
                try:
                    left = tmp_unmapped_file_name.replace( '.fastq', '_1.fastq' )
                    right = tmp_unmapped_file_name.replace( '.fastq', '_2.fastq' )
                    shutil.move( left, options.output_unmapped_reads_l )
                    shutil.move( right, options.output_unmapped_reads_r )
                except Exception, e:
                    sys.stdout.write( 'Error producing the unmapped output file.\n' )
            # check that there are results in the output file
            if os.path.getsize( options.output ) == 0:
                raise Exception, 'The output file is empty, there may be an error with your input file or settings.'
        except Exception, e:
            stop_err( 'Error aligning sequence. ' + str( e ) )
    finally:
        # clean up temp dir
        if os.path.exists( tmp_index_dir ):
            shutil.rmtree( tmp_index_dir )
    stdout += 'Sequence file aligned.\n'
    sys.stdout.write( stdout )

if __name__ == "__main__":
    __main__()

4:bdd9fcdb398e

    -x, --indexSettings=x: Whether or not indexing options are to be set
    -H, --suppressHeader=H: Suppress header
    --do_not_build_index: Flag to specify that provided file is already indexed and to just use 'as is'
"""

import optparse
import os
import shutil
import subprocess
import sys
import tempfile

# Allow more than Sanger encoded variants
DEFAULT_ASCII_ENCODING = '--phred33-quals'
GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG = {'fastqsanger': '--phred33-quals', 'fastqillumina': '--phred64-quals', 'fastqsolexa': '--solexa-quals'}
# FIXME: Integer quality scores are supported only when the '--integer-quals' argument is specified to bowtie; this is not currently able to be set in the tool/wrapper/config


def stop_err( msg ):
    sys.exit('%s\n' % msg)


def __main__():
    parser = optparse.OptionParser()
    parser.add_option( '-t', '--threads', dest='threads', help='The number of threads to run' )
    parser.add_option( '-o', '--output', dest='output', help='The output file' )
    parser.add_option( '', '--output_unmapped_reads', dest='output_unmapped_reads', help='File name for unmapped reads (single-end)' )
    parser.add_option( '', '--output_unmapped_reads_l', dest='output_unmapped_reads_l', help='File name for unmapped reads (left, paired-end)' )

    else:
        colorspace = ''
    # index if necessary
    if options.genomeSource == 'history' and not options.do_not_build_index:
        # set up commands
        if options.index_settings == 'indexPreSet':
            indexing_cmds = '%s' % colorspace
        else:
            try:
                if options.iautoB and options.iautoB == 'set':
                    iautoB = '--noauto'

                    iseed = ''
                indexing_cmds = '%s %s %s %s %s %s %s --offrate %s %s %s %s %s %s' % \
                                ( iautoB, ipacked, ibmax, ibmaxdivn, idcv, inodc,
                                  inoref, options.ioffrate, iftab, intoa, iendian,
                                  iseed, colorspace )
            except ValueError as e:
                # clean up temp dir
                if os.path.exists( tmp_index_dir ):
                    shutil.rmtree( tmp_index_dir )
                stop_err( "Something is wrong with the indexing parameters and the indexing and alignment could not be run. Make sure you don't have any non-numeric values where they should be numeric.\n" + str( e ) )
        ref_file = tempfile.NamedTemporaryFile( dir=tmp_index_dir )

        ref_file.close()
        os.symlink( options.ref, ref_file_name )
        cmd1 = 'bowtie-build %s -f %s %s' % ( indexing_cmds, ref_file_name, ref_file_name )
        try:
            tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
            with open(tmp, 'w') as tmp_stderr:
                returncode = subprocess.call(args=cmd1, shell=True, cwd=tmp_index_dir, stderr=tmp_stderr.fileno())
            if returncode != 0:
                # get stderr, allowing for case where it's very large
                stderr = ''
                buffsize = 1048576
                with open(tmp, 'r') as tmp_stderr:
                    try:
                        while True:
                            stderr += tmp_stderr.read(buffsize)
                            if not stderr or len(stderr) % buffsize != 0:
                                break
                    except OverflowError:
                        pass
                raise Exception(stderr)
        except Exception as e:
            # clean up temp dir
            if os.path.exists( tmp_index_dir ):
                shutil.rmtree( tmp_index_dir )
            stop_err( 'Error indexing reference sequence\n' + str( e ) )
        stdout += 'File indexed. '

    else:
        mateOrient = ''
    quality_score_encoding = GALAXY_FORMAT_TO_QUALITY_SCORE_ENCODING_ARG.get( options.galaxy_input_format, DEFAULT_ASCII_ENCODING )
    if options.params == 'preSet':
        aligning_cmds = '-q %s %s -p %s -S %s %s %s ' % \
            ( maxInsert, mateOrient, options.threads, suppressHeader, colorspace, quality_score_encoding )
    else:
        try:
            if options.skip and int( options.skip ) > 0:
                skip = '-s %s' % options.skip
            else:

                trimH = ''
            if options.trimL and int( options.trimL ) > 0:
                trimL = '-3 %s' % options.trimL
            else:
                trimL = ''
            if options.maxMismatches and (options.maxMismatches == '0' or options.maxMismatches == '1' or
                                          options.maxMismatches == '2' or options.maxMismatches == '3'):
                maxMismatches = '-v %s' % options.maxMismatches
            else:
                maxMismatches = ''
            if options.mismatchSeed and (options.mismatchSeed == '0' or options.mismatchSeed == '1' or
                                         options.mismatchSeed == '2' or options.mismatchSeed == '3'):
                mismatchSeed = '-n %s' % options.mismatchSeed
            else:
                mismatchSeed = ''
            if options.mismatchQual and int( options.mismatchQual ) >= 1:
                mismatchQual = '-e %s' % options.mismatchQual

                              maxAlignAttempt, forwardAlign, reverseAlign, maxBacktracks,
                              tryHard, valAlign, allValAligns, suppressAlign, best,
                              strata, offrate, seed, snpphred, snpfrac, keepends,
                              output_unmapped_reads, output_suppressed_reads,
                              quality_score_encoding )
        except ValueError as e:
            # clean up temp dir
            if os.path.exists( tmp_index_dir ):
                shutil.rmtree( tmp_index_dir )
            stop_err( 'Something is wrong with the alignment parameters and the alignment could not be run\n' + str( e ) )
    try:

                cmd2 = 'bowtie %s %s -1 %s -2 %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.input2, options.output )
            else:
                cmd2 = 'bowtie %s %s %s > %s' % ( aligning_cmds, ref_file_name, options.input1, options.output )
            # align
            tmp = tempfile.NamedTemporaryFile( dir=tmp_index_dir ).name
            with open(tmp, 'w') as tmp_stderr:
                returncode = subprocess.call(args=cmd2, shell=True, cwd=tmp_index_dir, stderr=tmp_stderr.fileno())
            # get stderr, allowing for case where it's very large
            stderr = ''
            buffsize = 1048576
            with open(tmp, 'r') as tmp_stderr:
                try:
                    while True:
                        stderr += tmp_stderr.read(buffsize)
                        if not stderr or len(stderr) % buffsize != 0:
                            break
                except OverflowError:
                    pass
            if returncode != 0:
                raise Exception(stderr)
            elif options.output_mapping_stats is not None:
                # Write stderr (containing the mapping statistics) to a named file
                with open(options.output_mapping_stats, 'w') as mapping_stats:
                    mapping_stats.write( stderr )
            # get suppressed and unmapped reads output files in place if appropriate
            if options.paired == 'paired' and tmp_suppressed_file_name and \
                    options.output_suppressed_reads_l and options.output_suppressed_reads_r:
                try:
                    left = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
                    right = tmp_suppressed_file_name.replace( '.fastq', '_1.fastq' )
                    shutil.move( left, options.output_suppressed_reads_l )
                    shutil.move( right, options.output_suppressed_reads_r )
                except Exception as e:
                    sys.stdout.write( 'Error producing the suppressed output file.\n' )
            if options.paired == 'paired' and tmp_unmapped_file_name and \
                    options.output_unmapped_reads_l and options.output_unmapped_reads_r:
                try:
                    left = tmp_unmapped_file_name.replace( '.fastq', '_1.fastq' )
                    right = tmp_unmapped_file_name.replace( '.fastq', '_2.fastq' )
                    shutil.move( left, options.output_unmapped_reads_l )
                    shutil.move( right, options.output_unmapped_reads_r )
                except Exception as e:
                    sys.stdout.write( 'Error producing the unmapped output file.\n' )
            # check that there are results in the output file
            if os.path.getsize( options.output ) == 0:
                raise Exception('The output file is empty, there may be an error with your input file or settings.')
        except Exception as e:
            stop_err( 'Error aligning sequence. ' + str( e ) )
    finally:
        # clean up temp dir
        if os.path.exists( tmp_index_dir ):
            shutil.rmtree( tmp_index_dir )
    stdout += 'Sequence file aligned.\n'
    sys.stdout.write( stdout )


if __name__ == "__main__":
    __main__()