def buildSelectStatementfromPed(filter_type, pedfile, template): '''Build a select statement from a template and a pedigree file''' pedigree = csv.DictReader( IOTools.open_file(pedfile), delimiter='\t', fieldnames=['family', 'sample', 'father', 'mother', 'sex', 'status']) affecteds = [] unaffecteds = [] parents = [] select = None # loop over pedigree file and establish relationships for row in pedigree: if row['status'] == '2': if filter_type == "denovo": father = row['father'] mother = row['mother'] proband = row['sample'] elif filter_type == "dominant" or filter_type == "recessive": affecteds += [row['sample']] if filter_type == "recessive": parents += [row['father'], row['mother']] if row['status'] == '1': if filter_type == "dominant": unaffecteds += [row['sample']] elif filter_type == "recessive": if row['sample'] not in parents: unaffecteds += [row['sample']] # Build select statement from template if filter_type == "denovo": select = template.replace("father", father) select = select.replace("mother", mother) select = select.replace("proband", proband) elif filter_type == "dominant": affecteds_exp = '").getPL().1==0&&vc.getGenotype("'.join(affecteds) if len(unaffecteds) == 0: unaffecteds_exp = '' else: unaffecteds_exp = '&&vc.getGenotype("' + \ ('").isHomRef()&&vc.getGenotype("'.join(unaffecteds)) + \ '").isHomRef()' select = template.replace("affecteds_exp", affecteds_exp) select = select.replace("unaffecteds_exp", unaffecteds_exp) elif filter_type == "recessive": affecteds_exp = '").getPL().2==0&&vc.getGenotype("'.join(affecteds) unaffecteds_exp = '").getPL().2!=0&&vc.getGenotype("'.join(unaffecteds) if len(parents) == 0: parents_exp = '' else: parents_exp = '&&vc.getGenotype("' + \ ('").getPL().1==0&&vc.getGenotype("'.join(parents)) + \ '").getPL().1==0' select = template.replace("affecteds_exp", affecteds_exp) select = select.replace("unaffecteds_exp", unaffecteds_exp) select = select.replace("parents_exp", parents_exp) return select
def main(argv=None): """script main. parses command line options in sys.argv, unless *argv* is given. """ if not argv: argv = sys.argv # setup command line parser parser = E.OptionParser(version="%prog version: $Id", usage=globals()["__doc__"]) parser.add_option("-u", "--ucsc-genome", dest="ucsc_genome", type="string", help="UCSC genome identifier [default=%default].") parser.add_option("-g", "--genome-file", dest="genome_file", type="string", help="filename with genome [default=%default].") parser.add_option("--extend", dest="extension", type="int", help="extend tags by this number of bases " "[default=%default].") parser.add_option("--shift-size", dest="shift", type="int", help="shift tags by this number of bases " "[default=%default].") parser.add_option("--window-size", dest="window_size", type="int", help="window size to be used in the analysis" "[default=%default].") parser.add_option("--saturation-iterations", dest="saturation_iterations", type="int", help="iterations for saturation analysis " "[default=%default].") parser.add_option("-t", "--toolset", dest="toolset", type="choice", action="append", choices=("saturation", "coverage", "enrichment", "dmr", "rms", "rpm", "all", "convert"), help="actions to perform [default=%default].") parser.add_option("-w", "--bigwig-file", dest="bigwig", action="store_true", help="store wig files as bigwig files - requires a " "genome file [default=%default]") parser.add_option("--treatment", dest="treatment_files", type="string", action="append", help="BAM files for treatment. At least one is required " "[%default]") parser.add_option("--control", dest="control_files", type="string", action="append", help="BAM files for control for differential " "methylation analysis. Optional [%default].") parser.add_option("--input", dest="input_files", type="string", action="append", help="BAM files for input correction. " "Optional [%default].") parser.add_option("--is-not-medip", dest="is_medip", action="store_false", help="data is not MeDIP data and is not expected " "to fit the calibration model. No CpG " "density normalized rms data is computed" "[default=%default].") parser.add_option("--output-rdata", dest="output_rdata", action="store_true", help="in dmr analysis, write R session to file. " "The file name " "is given by --ouptut-filename-pattern [%default].") parser.add_option("--rdata-file", dest="input_rdata", type="string", help="in dmr analysis, read saved R session from " "file. This can be used to apply different " "filters [%default]") parser.add_option("--fdr-threshold", dest="fdr_threshold", type="float", help="FDR threshold to apply for selecting DMR " "[default=%default].") parser.add_option("--fdr-method", dest="fdr_method", type="choice", choices=("bonferroni", "BH", "holm", "hochberg", "hommel", "BY", "fdr", "none"), help="FDR method to apply for selecting DMR " "[default=%default].") parser.add_option("--bwa", dest="bwa", action="store_true", help="alignment generated with bwa" "[default=%default].") parser.add_option("--unique", dest="unique", type="float", help="Threshold p-value to determine which read pile\ ups are the result of PCR overamplification" "[default=%default].") parser.add_option("--chroms", dest="chroms", type="str", help="Comma delimited list of chromosomes to include" "[default=%default].") parser.set_defaults(input_format="bam", ucsc_genome="Hsapiens.UCSC.hg19", genome_file=None, extend=0, shift=0, window_size=300, saturation_iterations=10, toolset=[], bigwig=False, treatment_files=[], control_files=[], input_files=[], output_rdata=False, input_rdata=None, is_medip=True, fdr_threshold=0.1, fdr_method="BH", bwa=False, unique=0.001, chroms=None) # add common options (-h/--help, ...) and parse command line (options, args) = E.start(parser, argv=argv, add_output_options=True) if "convert" in options.toolset: results = [] for line in CSV.DictReader(options.stdin, dialect="excel-tab"): if line['edgeR.p.value'] == "NA": continue # assumes only a single treatment/control treatment_name = options.treatment_files[0] control_name = options.control_files[0] status = "OK" try: results.append( Expression.GeneExpressionResult._make(( "%s:%i-%i" % (line['chr'], int(line['start']), int(line['stop'])), treatment_name, float(line['MSets1.rpkm.mean']), 0, control_name, float(line['MSets2.rpkm.mean']), 0, float(line['edgeR.p.value']), float(line['edgeR.adj.p.value']), float(line['edgeR.logFC']), math.pow(2.0, float(line['edgeR.logFC'])), float(line['edgeR.logFC']), # no transform ["0", "1"][float(line['edgeR.adj.p.value']) < options.fdr_threshold], status))) except ValueError as msg: raise ValueError("parsing error %s in line: %s" % (msg, line)) Expression.writeExpressionResults(options.stdout, results) return if len(options.treatment_files) < 1: raise ValueError("please specify a filename with sample data") if options.bigwig and not options.genome_file: raise ValueError("please provide a genome file when outputting bigwig") if options.genome_file: fasta = IndexedFasta.IndexedFasta(options.genome_file) contig_sizes = fasta.getContigSizes() if len(options.toolset) == 0: options.toolset = ["all"] do_all = "all" in options.toolset if options.chroms is None: chrstring = "" else: chroms = options.chroms.split(",") chrstring = ' chr.select=c(\"%s\"), ' % '\",\"'.join(chroms) # load MEDIPS R.library('MEDIPS') genome_file = 'BSgenome.%s' % options.ucsc_genome R.library(genome_file) window_size = options.window_size extend = options.extend shift = options.shift saturation_iterations = options.saturation_iterations uniq = float(options.unique) if options.bwa is True: BWA = "TRUE" else: BWA = "FALSE" if "saturation" in options.toolset or do_all: E.info("saturation analysis") for fn in options.treatment_files + options.control_files: paired = isPaired(fn) R('''sr = MEDIPS.saturation( file='%(fn)s', BSgenome='%(genome_file)s', shift=%(shift)i, extend=%(extend)i, window_size=%(window_size)i, uniq=%(uniq)s, nit = %(saturation_iterations)i, paired = %(paired)s, bwa = %(BWA)s, %(chrstring)s nrit = 1)''' % locals()) R.png(E.get_output_file("%s_saturation.png" % fn)) R('''MEDIPS.plotSaturation(sr)''') R('''dev.off()''') R('''write.table(sr$estimation, file ='%s', sep='\t')''' % E.get_output_file("%s_saturation_estimation.tsv" % fn)) outfile = IOTools.open_file( E.get_output_file("%s_saturation.tsv" % fn), "w") outfile.write("category\tvalues\n") outfile.write("estimated_correlation\t%s\n" % ",".join(["%f" % x for x in R('''sr$maxEstCor''')])) outfile.write("true_correlation\t%s\n" % ",".join(["%f" % x for x in R('''sr$maxTruCor''')])) outfile.write("nreads\t%s\n" % ",".join(["%i" % x for x in R('''sr$numberReads''')])) outfile.close() if "coverage" in options.toolset or do_all: E.info("CpG coverage analysis") for fn in options.treatment_files + options.control_files: paired = isPaired(fn) R('''cr = MEDIPS.seqCoverage( file='%(fn)s', BSgenome='%(genome_file)s', pattern='CG', shift=%(shift)i, extend=%(extend)i, paired=%(paired)s, bwa=%(BWA)s, %(chrstring)s uniq=%(uniq)s)''' % locals()) R.png(E.get_output_file("%s_cpg_coverage_pie.png" % fn)) R('''MEDIPS.plotSeqCoverage(seqCoverageObj=cr, type = "pie", cov.level = c(0, 1, 2, 3, 4, 5))''') R('''dev.off()''') R.png(E.get_output_file("%s_cpg_coverage_hist.png" % fn)) R('''MEDIPS.plotSeqCoverage(seqCoverageObj=cr, type = "hist", t=15)''') R('''dev.off()''') # note: this file is large R('''write.table(cr$cov.res, file=gzfile('%s','w'), sep='\t')''' % E.get_output_file("%s_saturation_coveredpos.tsv.gz" % fn)) if 'enrichment' in options.toolset or do_all: E.info("CpG enrichment analysis") outfile = IOTools.open_file(E.get_output_file("enrichment.tsv.gz"), "w") slotnames = (("regions.CG", "regions_CG", "%i"), ("regions.C", "regions_C", "%s"), ("regions.G", "regions_G", "%f"), ("regions.relH", "regions_relH", "%i"), ("regions.GoGe", "regions_GoGe", "%i"), ("genome.CG", "genome_CG", "%s"), ("genome.C", "genome_C", "%s"), ("genome.G", "genome_G", "%i"), ("genome.relH", "genome_relH", "%i"), ("enrichment.score.relH", "enrichment_relH", "%s"), ("enrichment.score.GoGe", "enrichment_GoGe", "%s")) outfile.write("\t".join(['sample'] + [x[1] for x in slotnames]) + "\n") for fn in options.treatment_files + options.control_files: paired = isPaired(fn) R('''ce = MEDIPS.CpGenrich( file='%(fn)s', BSgenome='%(genome_file)s', shift=%(shift)i, extend=%(extend)i, paired=%(paired)s, bwa=%(BWA)s, %(chrstring)s uniq=%(uniq)s)''' % locals()) outfile.write("%s" % fn) for slotname, label, pattern in slotnames: value = tuple(R('''ce$%s''' % slotname)) if len(value) == 0: value = "" outfile.write("\t%s" % pattern % value[0]) outfile.write("\n") outfile.close() if options.input_rdata: E.info("reading R session info from '%s'" % options.input_rdata) R('''load('%s')''' % options.input_rdata) else: if "dmr" in options.toolset or "correlation" in options.toolset \ or do_all: # build four sets for x, fn in enumerate(options.treatment_files): paired = isPaired(fn) E.info("loading '%s'" % fn) R('''treatment_R%(x)i = MEDIPS.createSet( file='%(fn)s', BSgenome='%(genome_file)s', shift=%(shift)i, extend=%(extend)i, window_size=%(window_size)i, paired=%(paired)s, bwa=%(BWA)s, %(chrstring)s uniq=%(uniq)s)''' % locals()) R('''treatment_set = c(%s)''' % ",".join([ "treatment_R%i" % x for x in range(len(options.treatment_files)) ])) if options.control_files: for x, fn in enumerate(options.control_files): paired = isPaired(fn) E.info("loading '%s'" % fn) R('''control_R%(x)i = MEDIPS.createSet( file='%(fn)s', BSgenome='%(genome_file)s', shift=%(shift)i, extend=%(extend)i, window_size=%(window_size)i, paired=%(paired)s, bwa=%(BWA)s, %(chrstring)s uniq=%(uniq)s)''' % locals()) R('''control_set = c(%s)''' % ",".join([ "control_R%i" % x for x in range(len(options.control_files)) ])) # build coupling vector R('''CS = MEDIPS.couplingVector(pattern="CG", refObj = treatment_set[[1]])''') if "correlation" in options.toolset or do_all: R('''cor.matrix = MEDIPS.correlation( c(treatment_set, control_set))''') R('''write.table(cor.matrix, file='%s', sep="\t")''' % E.get_output_file("correlation")) if "dmr" in options.toolset or do_all: # Data that does not fit the model causes # "Error in 1:max_signal_index : argument of length 0" # The advice is to set MeDIP=FALSE # See: http://comments.gmane.org/ # gmane.science.biology.informatics.conductor/52319 if options.is_medip: medip = "TRUE" else: medip = "FALSE" fdr_method = options.fdr_method E.info("applying test for differential methylation") R('''meth = MEDIPS.meth( MSet1 = treatment_set, MSet2 = control_set, CSet = CS, ISet1 = NULL, ISet2 = NULL, p.adj = "%(fdr_method)s", diff.method = "edgeR", MeDIP = %(medip)s, CNV = F, minRowSum = 1)''' % locals()) # Note: several Gb in size # Output full methylation data table R('''write.table(meth, file=gzfile('%s', 'w'), sep="\t", row.names=F, quote=F)''' % E.get_output_file("data.tsv.gz")) # save R session if options.output_rdata: R('''save.image(file='%s', safe=FALSE)''' % E.get_output_file("session.RData")) # DMR analysis - test for windows and output if "dmr" in options.toolset: E.info("selecting differentially methylated windows") # test windows for differential methylation fdr_threshold = options.fdr_threshold R('''tested = MEDIPS.selectSig(meth, adj=T, ratio=NULL, p.value=%(fdr_threshold)f, bg.counts=NULL, CNV=F)''' % locals()) R('''write.table(tested, file=gzfile('%s', 'w'), sep="\t", quote=F)''' % E.get_output_file("significant_windows.gz")) # select gain and merge adjacent windows try: R('''gain = tested[which(tested[, grep("logFC", colnames(tested))] > 0),]; gain_merged = MEDIPS.mergeFrames(frames=gain, distance=1)''') E.info('gain output: %s, merged: %s' % (str(R('''dim(gain)''')), str(R('''dim(gain_merged)''')))) R('''of=gzfile('%s', 'w'); write.table(gain_merged, file=of, sep="\t", quote=F, row.names=FALSE, col.names=FALSE); close(of)''' % E.get_output_file("gain.bed.gz")) except rpy2.rinterface.RRuntimeError as msg: E.warn("could not compute gain windows: msg=%s" % msg) # select loss and merge adjacent windows try: R('''loss = tested[which(tested[, grep("logFC", colnames(tested))] < 0),]; loss_merged = MEDIPS.mergeFrames(frames=loss, distance=1)''') E.info('loss output: %s, merged: %s' % (str(R('''dim(loss)''')), str(R('''dim(loss_merged)''')))) R('''of=gzfile('%s', 'w'); write.table(loss_merged, file=of, sep="\t", quote=F, row.names=F, col.names=F); close(of)''' % E.get_output_file("loss.bed.gz")) except rpy2.rinterface.RRuntimeError as msg: E.warn("could not compute loss windows: msg=%s" % msg) # if "rpm" in options.toolset or do_all: # outputfile = E.get_output_file("rpm.wig") # R('''MEDIPS.exportWIG(file = '%(outputfile)s', # data = CONTROL.SET, raw = T, descr = "rpm")''' % # locals()) # if options.bigwig: # bigwig(outputfile, contig_sizes) # else: # compress(outputfile) # if "rms" in options.toolset or do_all: # outputfile = E.get_output_file("rms.wig") # R('''MEDIPS.exportWIG(file = '%(outputfile)s', # data = CONTROL.SET, raw = F, descr = "rms")''' % # locals()) # if options.bigwig: # bigwig(outputfile, contig_sizes) # else: # compress(outputfile) # write footer and output benchmark information. E.stop()