def __init__(self):
self.blastTools = BlastTools()
self.verbose = False
class ProcessVCF:
'''
@summary:
Tools set for dealing with variant call format (vcf) Files.
Specifically this tools set does the following
1. Find expanding regions of variants with a designated distance across all samples
2. Combine variants in each sample into a report object to be used for machine learning analysis of genotype to phenotype.
3. Create a distance matrix using given variants between given strains.
'''
def __init__(self):
self.blastTools = BlastTools()
self.verbose = False
def writeToLog(self,stringValue,logFileH,verbose=False):
if logFileH == None:
return None
try:
if verbose: print stringValue
logFileH.write(stringValue+"\n")
except:
print "failed to write to log file [%s]" % (logFileH)
return None
def replaceSeqTarget(self,seq,newSeq,loc):
prefix = seq[:loc]
post = seq[loc+len(newSeq):]
result = prefix + newSeq + post
return result
def findTargets(self,targetMap,feature,minQuality=10,logFile=None):
'''
@var
@summary:
Find variant calls in proximity to given feature.
'''
start = feature.location.start.position
end = feature.location.end.position
locations = targetMap.keys()
matches = filter(lambda x: start<=float(x)<=end,locations)
querySeq = feature.qualifiers["query"]
subjectSeq = feature.qualifiers["subject"]
refSeq = Seq("_"*len(subjectSeq))
readSeq = Seq("_"*len(subjectSeq))
qualityValues = []
chrom = ''
dCount = 0
for loc in matches:
floc = float(loc)
seqStart = int(floc-start)
target = targetMap[loc]
chrom = target["CHROM"]
ref = target["REF"]
alts = target["ALT"].split(",")
quality = target["QUAL"]
qualityValues.append((seqStart,quality))
if float(quality) < minQuality:
print "(failed) Feature [%s] (%s <-> %s) ===> %s" % (feature.id,start,end,qualityValues)
else:
dCount += 1
refSeq = self.replaceSeqTarget(refSeq,ref,seqStart)
for alt in alts:
readSeq = self.replaceSeqTarget(readSeq,alt,seqStart)
result = {}
result["name"] = feature.id
result["chrom"] = chrom
result["start"] = start
result["end"] = end
result["quality"] = qualityValues
result["query"] = querySeq
result["subject"] = subjectSeq
result["refSeq"] = refSeq
result["alt-read"] = readSeq
return result
def annotateAlignment(self,targetMap,featuresArray,idTag):
readLogName = "read_log_%s.txt" % (idTag)
logFP = open(readLogName,"w")
result = {}
for features in featuresArray:
for feature in features:
id = feature.id
targets = self.findTargets(targetMap, feature, minQuality= 10, logFile = logFP)
result[id] = targets
return result
def alignVCF(self,targetFile,vcfFile,idTag):
'''
@return: ReportObect
@summary:
Check alignment to annotated Genomic sequence using BLAST.
'''
readRecord = parseVCFFile(vcfFile)
targetRecords= SeqIO.parse(open(targetFile), "fasta")
targetRecords = list(targetRecords)
print "Processing [%s] records" % (len(targetRecords))
if verbose: print "blasting sequences"
self.blastTools.verbose = verbose
blastedFeatures = self.blastTools.seqBlastToFeatures(blastDB, blastExe, targetFile, blastType = "blastn",scoreMin = 1e-5)
if verbose: print "finished blasting locations"
alignmentReport = self.annotateAlignment(readRecord, blastedFeatures,idTag)
return alignmentReport