def getCons(path, tfName):
for infile in glob.glob(os.path.join(path, "*.wigFix")):
(wigpath,wigfilename) = os.path.split(infile)
chrom = wigfilename.split('.')[0]
consName = '_'.join(wigfilename.split('.')[1:3])
#print chrom, tfName, consName
with open(infile,'rt') as wigFile:
wig = csv.reader(wigFile,delimiter='\t')
stepDict, startDict, valuesDict = countWig.getFixStart(wig,consName)#'phyloP30wayEuarchontoglires')
start = startDict[consName][chrom]
arrayDict = countWig.buildFixHist(chrom,stepDict,startDict,valuesDict,consName)
cursor = mcollection.find({"tf_name": tfName,
"motif_genomic_regions_info.chr": chrom})
for test in cursor:
motifStart, motifEnd = test["motif_genomic_regions_info"]["start"], test["motif_genomic_regions_info"]["end"]
avg = 0
#print motifStart, motifEnd
startlist = [start[i] for i in xrange(len(start)-1) if (motifStart > start[i] and motifStart < start[i+1]) or (motifEnd > start[i] and motifEnd < start[i+1])]
if len(startlist) > 0:
print startlist
#print start[-1]
startlist.append(start[-1])
for i in xrange(len(startlist)):
#print arrayDict[start[i]]
# if avg != 0 and motifEnd < start[i]:
# break ##fall into range and break out
if avg != 0:
if motifEnd > startlist[i]:##cases of partial overlap need to renormalize over two fragments
xs, xvals, sums = arrayDict[startlist[i]]
avg = avg * (startlist[i] - motifStart) + (countWig.queryHist(xs,
xvals, sums, motifStart, motifEnd)[0] * (motifEnd - startlist[i] + 1)) / (motifEnd - motifStart + 1)
else:
break
elif avg == 0:
xs, xvals, sums = arrayDict[startlist[i]]
avg = countWig.queryHist(xs, xvals, sums, motifStart, motifEnd)[0]
if avg > 0:
print avg, motifStart, motifEnd
mcollection.update({"_id":test["_id"]},{"$set":{"motif_cons_info":{consName: avg}}}, upsert = True)
else:
mcollection.update({"_id":test["_id"]},{"$set":{"motif_cons_info":{consName: avg}}}, upsert = True)
#mcollection.save(test)
return 0
