def updateMap(path, db, motifChrom="chr17", window=100):
"""update mapability scores from bedGraph files"""
mcollection = db["hg19"+motifChrom]
for infile in glob.glob(os.path.join(path, "*"+motifChrom+".bedGraph")):##better in bedGraph
(mappath,mapfilename) = os.path.split(infile)
expName = mapfilename.split(motifChrom)[0]
print 'updating', expName, motifChrom
#print expName
with open(infile,'rt') as bedGraphFile:
#bedGraph = csv.reader(bedGraphFile, delimiter = '\t')
bbFile = os.path.join(mappath,expName+motifChrom+'.bb')
if not os.path.isfile(bbFile):
countBed.compressBed4(bedGraphFile, expName, bbFile)
coordDict, valuesDict = countBed.getBinBedCoord(bbFile, expName)
arrayDict=defaultdict(list)
cursor = mcollection.find()#{"tf_name":tfName})
for test in cursor:
motifStart, motifEnd = test["genomic_region"]["start"], \
test["genomic_region"]["end"]
if not motifChrom in arrayDict:
arrayDict[motifChrom] = countBed.buildBedHist(motifChrom, coordDict, valuesDict, expName)
xs, xvals, sums = arrayDict[motifChrom]
avg = countBed.queryHist(xs, xvals, sums, motifStart-window, motifEnd+window)[0]
#if avg != 0:
#print motifChrom, motifStart, motifEnd
#print avg
mcollection.update({"_id":test["_id"]},{"$set":{"map."+expName:avg}}, upsert = True)
return 0
def updateMap(path, tfName, window):
"""update mapability scores from bedGraph files"""
for infile in glob.glob(os.path.join(path, "*.bedGraph")):##better in bedGraph
(mappath,mapfilename) = os.path.split(infile)
expName = mapfilename.split('.')[0]
#print expName
with open(infile,'rt') as bedGraphFile:
bedGraph = csv.reader(bedGraphFile, delimiter = '\t')
coordDict, valuesDict = countBed.getBed4Coord(bedGraph, expName)
arrayDict=defaultdict(list)
cursor = mcollection.find({"tf_name":tfName})
for test in cursor:
motifChrom, motifStart, motifEnd = test["motif_genomic_regions_info"]["chr"], \
test["motif_genomic_regions_info"]["start"], \
test["motif_genomic_regions_info"]["end"]
if not motifChrom in arrayDict:
arrayDict[motifChrom] = countBed.buildBedHist(motifChrom, coordDict, valuesDict, expName)
xs, xvals, sums = arrayDict[motifChrom]
avg, size = countBed.queryHist(xs, xvals, sums, motifStart-window, motifEnd+window)
if avg != 0:
print motifChrom, motifStart, motifEnd
print avg
mcollection.update({"_id":test["_id"]},{"$set":{"motif_mapability_info":{"score":{expName:avg}}}}, upsert = True)
return 0
def updateMap(inpath,infile,outpath,motifChrom,seqlen,window=100):
"""update mapability scores from bedGraph files"""
motifdir = os.path.join(outpath,"bedMotifs")
mapdir = os.path.join(outpath,"mapMotifs")
if not os.path.isdir(motifdir):
print "Error: path-to-motif-bed-files invalid, please specify a valid outpath to store all calculated scores."
sys.exit()
if not os.path.isdir(mapdir):
os.mkdir(mapdir)
mapinfile = os.path.join(inpath, infile)##better in per chromosome bedGraph files
(mappath,mapfilename) = os.path.split(mapinfile)
expName = mapfilename.split(motifChrom)[0]
print 'updating', expName, motifChrom
with bz2.BZ2File(mapinfile,'r') as bedGraphFile:
bbFile = os.path.join(inpath,expName+motifChrom+'.bb')
if not os.path.isfile(bbFile):
countBed.compressBed4(bedGraphFile, expName, bbFile)
coordDict, valuesDict = countBed.getBinBedCoord(bbFile, expName)
arrayDict=defaultdict(list)
for bedfile in glob.glob(os.path.join(motifdir,"*"+motifChrom+".bed.gz")):
(filepath,filename) = os.path.split(bedfile)
tfname = filename.split(motifChrom)[0]
gcoordsfile = gzip.open(bedfile,'r')
gcoords = csv.reader(gcoordsfile, delimiter='\t')
mapfile = gzip.open(os.path.join(mapdir,tfname+motifChrom+'map'+str(seqlen)+'.txt.gz'),'w')
writer = csv.writer(mapfile, delimiter='\t')
for test in gcoords:
motifStart, motifEnd = int(test[1]), int(test[2])
if not motifChrom in arrayDict:
arrayDict[motifChrom] = countBed.buildBedHist(motifChrom, coordDict, valuesDict, expName)
xs, xvals, sums = arrayDict[motifChrom]
avg = countBed.queryHist(xs, xvals, sums, motifStart-window, motifEnd+window)[0]
row = [avg]
writer.writerows([row])
mapfile.close()
return 0
