def GetTotalNumbers(realKmerDict, posFile, negFile, pssmList, pwm, predictedMotifs, predictedKmers=None): PosSeqDict = SeqGenUtils.fasta_read(posFile); NegSeqDict = SeqGenUtils.fasta_read(negFile); numPosTP = numPosFP = numPosFN = 0; numNegTP = numNegFP = numNegFP = 0; #print "Predicted KMER DICT: ", predictedKmers; numPosTP, numPosFP, numPosFN = getTotalNumbersForSeqDict(PosSeqDict, realKmerDict, pssmList, pwm, predictedMotifs, predictedKmers); numNegTP, numNegFP, numNegFN = getTotalNumbersForSeqDict(NegSeqDict, realKmerDict, pssmList, pwm, predictedMotifs, predictedKmers); #print "Pos File: TP: ", str(numPosTP), ", FP: ", numPosFP, ", FN: ", numPosFN #print "Neg File: TP: ", str(numNegTP), ", FP: ", numNegFP, ", FN: ", numNegFN return (numPosTP + numNegTP), (numPosFP + numNegFP) , (numPosFN + numNegFN);
def GetDistribListForDirectory(directory, filesToFind): distribList = list(); for seqFile in SeqGenUtils.findFiles(directory, filesToFind): seqs, gc_list, fg_lengths = shuffle_utils.get_seqs(seqFile) dinuc_distrib = shuffle_utils.compute_dinuc_distrib(seqs, True) distribList.append(dinuc_distrib) return distribList;
def GetKmersFromStructureFile(structureFile, numSeqsWithSignal): structureDict = SeqGenUtils.fasta_read(structureFile); structureDict = SeqGenUtils.ChangeUsToTs(structureDict); originalKmers = structureDict.values(); updatedKmers = CreateLengthMatchingKmers(originalKmers, numSeqsWithSignal) return updatedKmers;
def ComputeNgramFrequencyAndProbability(seqFile, nLen):
seqDict = SeqGenUtils.fasta_read(seqFile);
nGramCombinedList = list();
total_seq_length = 0;
for header, sequence in seqDict.iteritems():
ngram_list = getNgramListForSeq(sequence, nLen)
if len(ngram_list) > 0:
nGramCombinedList.append(ngram_list)
total_seq_length = total_seq_length + len(sequence)
else:
print "Found empty sequence for ", header;
ngram_freq = getNGramFreqForCombinedList(nGramCombinedList, nLen);
ngram_freq = dict(ngram_freq)
ngram_prob = {}
for ngram, frequency in ngram_freq.iteritems():
ngram_prob[ngram] = round(frequency/total_seq_length, 4);
# print str(nLen) + "-" + "gram frequences: \n", ngram_freq;
#print str(nLen) + "-" + "gram probabilities: \n", ngram_prob;
#print "Total Nucleotides: ", str(total_seq_length)
return ngram_freq, ngram_prob;
def GetRandomSequence(self, seqLen):
seq = ""
for count in range(seqLen):
seq += SeqGenUtils.weightedchoice([("C", self.cPercent),
("G", self.gPercent),
("A", self.aPercent),
("T", self.tPercent)])
return seq
def test_DictFromFasta(self): structureAlignmentFile = "/projects/bhandare/workspace/PySG/src/resources/RF000037.fa" structureDict = SeqGenUtils.fasta_read(structureAlignmentFile); self.assertEqual(len(structureDict), 62); for key, value in structureDict.iteritems(): condition = 'T' in value; self.assertFalse(condition); structureDict = SeqGenUtils.ChangeUsToTs(structureDict); self.assertEqual(len(structureDict), 62); for key, value in structureDict.iteritems(): condition = 'T' in value; self.assertTrue(condition);
def computeSequenceBasedDREMEResults(dremeFile, realCsvFile, posSeqFile, negSeqFile): totalPosTP = totalPosFP = totalPosFN = totalPosTN = 0; totalNegTP = totalNegFP = totalNegFN = totalNegTN = 0; posSeqDict = SeqGenUtils.fasta_read(posSeqFile); negSeqDict = SeqGenUtils.fasta_read(negSeqFile); realKmerDict = parseRealKmers.GetRealKmerDict(realCsvFile); pssmList = parseDreme.getPSSMListFromDremeFile(dremeFile) numPosTP, numPosFP, numPosFN, numPosTN = compareRealAndPredicted(realKmerDict, posSeqDict, pssmList, positive=True) numNegTP, numNegFP, numNegFN, numNegTN = compareRealAndPredicted(realKmerDict, negSeqDict, pssmList, False) print "Positive: TP: ", numPosTP, ", FP: ", numPosFP, ", FN: ", numPosFN, ", TN: ", numPosTN print "Negative: TP: ", numNegTP, ", FP: ", numNegFP, ", FN: ", numNegFN, ", TN: ", numNegTN totalPos = len(posSeqDict) totalNeg = len(negSeqDict) sensitivity, ppv = compareKmers.GetSensitivityAndPPV((numPosTP + numNegTP) , (numPosFP + numNegFP), (numPosFN + numNegFN)) accuracy = compareKmers.GetAccuracy( (numPosTP + numNegTP), (numPosTN + numNegTN), (totalPos + totalNeg) ) specificity = compareKmers.GetSpecificity( (numPosFP + numNegFP), totalNeg); print "Senitivity: ", sensitivity, ", PPV: ", ppv, ", Accuracy: ", accuracy, ", Specificity: ", specificity; return sensitivity, ppv;
def parseSubDirectories(resultDir, level=1):
gcContentMap = dict();
resultFileName = resultDir + resultDir[:-1] + "_GC_Content.out"
print "Result FileName: ", resultFileName
for signalFile in SeqGenUtils.findFiles(resultDir, "Signal*.fa"):
print "Signal File: ", signalFile;
expt_name = os.path.dirname(signalFile).split("/")[2]
gcContentValue, atContentValue = gcContent.getNucleotideComposition(signalFile)
print gcContentValue, atContentValue;
if expt_name in gcContentMap.keys():
gcContentMap[str(expt_name)].append(gcContentValue);
else:
gcContentMap[str(expt_name)] = [gcContentValue];
gcContentMeanStd = dict();
for key, value in gcContentMap.iteritems():
meanValue = np.mean(value);
varianceValue = np.var(value);
gcContentMeanStd[key] = [meanValue, varianceValue]
writeDictToFile(gcContentMeanStd, resultFileName);
motifCountDict[motif] = 1; if motifFound == False: totalMisMatches = totalMisMatches + 1; print "Total Mismatches: ", totalMisMatches; return motifMatchDict, motifCountDict; if __name__ == "__main__": import sys motifFile = sys.argv[1] faFile = sys.argv[2] motifList = getMotifList(motifFile) print motifList seqDict = SeqGenUtils.fasta_read(faFile) motifMatchDict, motifCountDict = findMotifInSequences(motifList, seqDict) singleMotif = 0 multipleMotif = 0; for header, value in motifMatchDict.iteritems(): if len(value) > 1: multipleMotif = multipleMotif + 1; if len(value) == 1: singleMotif = singleMotif + 1; print "Sequences that matches single motif: ", singleMotif; print "Sequences that matches multiple motifs: ", multipleMotif; print "Total Sequences: ", len(seqDict) # for item in sorted(motifMatchDict):
def writeFastaFileFromKmers(kmerList, prefix, outFileName): fastaFileLines = SeqGenUtils.createFastaFileFromKmers(kmerList, prefix); SeqGenUtils.writeFastaLinesToFile(fastaFileLines, outFileName);
def GetRandomSequence(self, seqLen):
seq=""
for count in range(seqLen):
seq+=SeqGenUtils.weightedchoice([("C", self.cPercent), ("G", self.gPercent), ("A",
self.aPercent), ("T", self.tPercent)]);
return seq;
kmerDict = parseKspectrum.FindKspectrumKmers(featureFile, numKmers) HuR_ReString = '[^-](\d+\.\d+)\,HuR_([ATGC]+)' TTP_ReString = '[^-](\d+\.\d+)\,TTP_([ATGC]+)' HuRKmerDict = parseKspectrum.FindRBPSpecificKmers(featureFile, HuR_ReString, numKmers); TTPKmerDict = parseKspectrum.FindRBPSpecificKmers(featureFile, TTP_ReString, numKmers); kmerFastaLines = addFastaHeaderToKmers(kmerDict, "FeatureKmer") HuRFastaLines = addFastaHeaderToKmers(HuRKmerDict, "HuR_Specific") TTPFastaLines = addFastaHeaderToKmers(TTPKmerDict, "TTP_Specific") filename_prefix = os.path.splitext(sequenceFile)[0]; filename_ext = os.path.splitext(sequenceFile)[1]; HuR_Filename = filename_prefix + "_HuR" + filename_ext; TTP_Filename = filename_prefix + "_TTP" + filename_ext; SeqGenUtils.writeFastaLinesToFile(kmerFastaLines, sequenceFile) SeqGenUtils.writeFastaLinesToFile(HuRFastaLines, HuR_Filename) SeqGenUtils.writeFastaLinesToFile(TTPFastaLines, TTP_Filename) # topKmerFile = sys.argv[1] # signalType = sys.argv[2] ## PWM or PFM # signalFile = sys.argv[3] # seqFile = sys.argv[4] # topKmers = int(sys.argv[5]) # kmerList = GetTopXKmers(topKmerFile, topKmers); # motif = GetMotifForSignal(signalType, signalFile); # seqDict = SeqGenUtils.fasta_read(seqFile); # FindBestScoreSeqForMotif(motif, seqDict, kmerList)
