gatk-3.8/python/Gelis2PopSNPs.py

import farm_commands
import os.path
import sys
from optparse import OptionParser
import string
import re
import glob
import picard_utils
import itertools

gatkPath = "~/dev/GenomeAnalysisTK/trunk/dist/GenomeAnalysisTK.jar"
ref = "/seq/references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta"
   
def geli2dbsnpFile(geli):
    root, flowcellDotlane, ext = picard_utils.splitPath(geli)
    return os.path.join(root, flowcellDotlane) + '.dbsnp_matches'

def SingleSampleGenotyperCmd(bam, geli, use2bp):
    naid = os.path.split(bam)[1].split(".")[0]
    metrics = geli + '.metrics'
    gatkPath = '/humgen/gsa-scr1/kiran/repositories/Sting/trunk/dist/GenomeAnalysisTK.jar'
    hapmapChip = '/home/radon01/andrewk/hapmap_1kg/gffs/' + naid + '.gff'
    hapmapChipStr = ' '.join(['--hapmap_chip', hapmapChip]) 
    if not os.path.exists(hapmapChip):
        print '*** warning, no hapmap chip resuls for', naid
        hapmapChipStr = ''

    targetList = '/home/radon01/depristo/work/1kg_pilot_evaluation/data/thousand_genomes_alpha_redesign.targets.interval_list'
    cmd = "java -ea -jar " + gatkPath + ' ' + ' '.join(['-T SingleSampleGenotyper', '-I', bam, '-L', targetList, '-R', ref, '-D', '/humgen/gsa-scr1/GATK_Data/dbsnp_129_hg18.rod', '-metout', metrics, '-varout', geli, '-geli -l INFO ']) + hapmapChipStr
    return cmd
 
def bams2geli(bams):
    def call1(bam):
        geli = os.path.splitext(bam)[0] + '.geli'
        jobid = 0
        if OPTIONS.useSSG:
            if not os.path.exists(geli + '.calls'):
                cmd = SingleSampleGenotyperCmd(bam, geli + '.calls', OPTIONS.useSSG2b)
        else:
            if not os.path.exists(geli):
                cmd = picard_utils.callGenotypesCmd( bam, geli, options = picard_utils.hybridSelectionExtraArgsForCalling())
        jobid = farm_commands.cmd(cmd, OPTIONS.farmQueue, just_print_commands = OPTIONS.dry )
        return geli, jobid
    calls = map(call1, bams)
    return map(lambda x: x[0], calls), map(lambda x: x[1], calls)        
   
def gelis2gelisText( gelis ):
    def geli2geliText( maybeGeli ):
        if os.path.splitext(maybeGeli)[1] == ".calls" :
            return maybeGeli
        else:
            return os.path.split(geli)[1] + '.calls'
            
    return map( geli2geliText, gelis)

   
def main():    
    global OPTIONS, ROOT

    usage = "usage: %prog lanes.list nIndividuals [options]"
    parser = OptionParser(usage=usage)
    parser.add_option("-q", "--farmQueue", dest="farmQueue",
                        type="string", default=None,
                        help="Farm queue to submit jobs to.  Leave blank for local processing")
    parser.add_option("-l", "--lod", dest="lod",
                        type="float", default=5,
                        help="minimum lod for calling a variant")
    parser.add_option("-k", "--column", dest="column",
                        type="int", default=1,
                        help="Column in the file with the bam or geli file path")
    parser.add_option("", "--dry", dest="dry",
                        action='store_true', default=False,
                        help="If provided, nothing actually gets run, just a dry run")
    parser.add_option("", "--ssg", dest="useSSG",
                        action='store_true', default=False,
                        help="If provided, we'll use the GATK SSG for genotyping")
    parser.add_option("", "--ssg2b", dest="useSSG2b",
                        action='store_true', default=False,
                        help="If provided, we'll use 2bp enabled GATK SSG ")
    parser.add_option("-o", "--output", dest="output",
                        type="string", default='/dev/stdout',
                        help="x")
                        
    (OPTIONS, args) = parser.parse_args()
    if len(args) != 2:
        parser.error("incorrect number of arguments: " + str(args))
    lines = [line.split() for line in open(args[0])]
    nIndividuals = int(args[1])
    
    outputFile = open(OPTIONS.output, 'w')
    print >> outputFile, '#', ' '.join(sys.argv)
    
    data = map( lambda x: x[OPTIONS.column-1], lines )
    if os.path.splitext(data[0])[1] == '.bam':
        gelis, jobids = bams2geli(data)
        if filter(lambda x: x <> 0, jobids) <> []:
            # there's still work to do
            sys.exit('Stopping.  Please rerun this program when the farm jobs are complete: ' + str(jobids))
            # TODO: Should add a wait here for all farm jobs to finish...
        print 'gelis', gelis
        print 'jobids', jobids
    else:
        gelis = map( lambda x: x[OPTIONS.column-1], lines )
        jobids = [None] * len(gelis)
    
    print 'Geli files'
    print gelis

    for geli, jobid in zip(gelis, jobids):
        dbsnpFile = geli2dbsnpFile(geli)
        if not os.path.exists(dbsnpFile):
            dbsnpCmd = picard_utils.CollectDbSnpMatchesCmd(geli, dbsnpFile, OPTIONS.lod)
            if jobid == 0: jobid = None
            farm_commands.cmd(dbsnpCmd, OPTIONS.farmQueue, just_print_commands = OPTIONS.dry, waitID = jobid)

    # TODO: Should add a wait here for all farm jobs to finish...

    # read in the dbSNP tracks
    nTotalSnps = 0
    nNovelSnps = 0
    for geli in gelis:
        root, flowcellDotlane, ext = picard_utils.splitPath(geli)
        #dbsnp_matches = os.path.join(root, flowcellDotlane) + '.dbsnp_matches'
        dbsnp_matches = geli2dbsnpFile(geli)
        print dbsnp_matches
        if os.path.exists(dbsnp_matches):
            TOTAL_SNPS, NOVEL_SNPS, PCT_DBSNP, NUM_IN_DB_SNP = picard_utils.read_dbsnp(dbsnp_matches)
            nTotalSnps += int(TOTAL_SNPS)
            nNovelSnps += int(NOVEL_SNPS)
            print >> outputFile, '# DATA:    ', flowcellDotlane, TOTAL_SNPS, NOVEL_SNPS, PCT_DBSNP, NUM_IN_DB_SNP, dbsnp_matches
    print  >> outputFile, '# DATA:    TOTAL SNP CALLS SUMMED ACROSS LANES, NOT ACCOUNT FOR IDENTITY', nTotalSnps
    print  >> outputFile, '# DATA:    NOVEL SNP CALLS SUMMED ACROSS LANES, NOT ACCOUNT FOR IDENTITY ', nNovelSnps
    print  >> outputFile, '# DATA:    AVERAGE DBSNP RATE ACROSS LANES %.2f' % (100.0 * float(nTotalSnps - nNovelSnps) / (max(nTotalSnps, 1)))

    # convert the geli's to text
    jobid = None
    variantsOut = gelis2gelisText( gelis ) 
    for geli, variantOut in zip(gelis, variantsOut):
        name = os.path.split(geli)[1]
        if not os.path.exists(variantOut):
            cmd = ("GeliToText.jar I=%s | awk '$1 !~ \"@\" && $1 !~ \"#Sequence\" && $0 !~ \"GeliToText\"' | awk '$7 > %f {print \"%s\" $0}' > %s" % ( geli, OPTIONS.lod, name, variantOut) )
            jobid = farm_commands.cmd(cmd, OPTIONS.farmQueue, just_print_commands = OPTIONS.dry)

    cmd = ("cat %s | sort -k 1 -k 2 -n > tmp.calls" % ( ' '.join(variantsOut) ) )
    jobid = farm_commands.cmd(cmd, OPTIONS.farmQueue, just_print_commands = OPTIONS.dry, waitID = jobid)

    sortedCallFile = 'all.sorted.calls'
    cmd = ("~/dev/GenomeAnalysisTK/trunk/perl/sortByRef.pl -k 1 tmp.calls ~/work/humanref/Homo_sapiens_assembly18.fasta.fai > %s"  % ( sortedCallFile ) )    
    jobid = farm_commands.cmd(cmd, OPTIONS.farmQueue, just_print_commands = OPTIONS.dry, waitID = jobid)
     
    sortedCalls = [line.split() for line in open(sortedCallFile)]
    aggregratedCalls = picard_utils.aggregateGeliCalls(sortedCalls)
    
    print >> outputFile, 'loc ref alt EM_alt_freq discovery_likelihood discovery_null discovery_prior discovery_lod EM_N n_ref n_het n_hom individuals'

    for snp in map( lambda x: picard_utils.aggregatedGeliCalls2SNP(x, nIndividuals), aggregratedCalls ):
        if snp == None: continue    # ignore bad calls
        #print snp
        #sharedCalls = list(sharedCallsGroup)
        #genotype = list(sharedCalls[0][5])
        print >> outputFile, '%s   %s %s %.6f -420.0 -420.0 0.000000 100.0 %d %d %d %d NA' % (snp.loc, snp.ref, snp.alt(), snp.q(), nIndividuals, snp.nRefGenotypes(), snp.nHetGenotypes(), snp.nHomVarGenotypes())
    outputFile.close()
    
                
if __name__ == "__main__":
    main()