def run(args):
logging.basicConfig(level=logging.DEBUG)
fin=IO.fopen(args.input,"r")
out=IO.fopen(args.output,"w")
bam=DBI.init(args.bam,"bam");
beds=[i for i in TableIO.parse(fin,"bed12")]
beds.sort()
bp=args.bp
print("mapped:{}".format(bam.mapped))
print("unmapped:{}".format(bam.unmapped))
data={}
for i,x in enumerate(iter_cluster(beds)):
print("{}\t{}:{}-{}".format(i+1,x["chr"],x["start"]+1,x["stop"]))
'''
cds=[z.cds() for z in x["beds"] if z.cds()]
utr3=[z.utr3() for z in x["beds"] if z.utr3()]
utr5=[z.utr5() for z in x["beds"] if z.utr5()]
'''
coords = [ up_down_coordinate(gene,args.bp,args.bp) for gene in x["beds"] ]
for j,y in enumerate(coords):
data[y.id]={}
data[y.id]["coord"]=y
data[y.id]["values"]=[0.0 for l in range(y.cdna_length())];
coord_beds = [ _translate(coord,bed) for coord,bed in itertools.izip(coords,x["beds"])]
for j,read in enumerate(bam.query(method="bam1",chr=x["chr"],start=x["start"]-args.bp,stop=x["stop"]+args.bp,strand=args.strand)):
NM=getNM(read) # number of hits
NC=0 # number of compatible
c_coords=[]
for k,coord in enumerate(coords):
if overlap(read,coord) and compatible(read,coord): # don't consider the reads extend out of coords.
NC+=1
c_coords.append(k)
for k,c in enumerate(c_coords):
coord=coords[c]
if read.start < coord.start or read.stop > coord.stop:
start=max(read.start,coord.start)
stop=min(read.stop,coord.stop)
read=read._slice(start,stop)
read_in_coord = _translate(coord,read)
for l in xrange(read_in_coord.start,read_in_coord.stop):
data[coord.id]["values"][l]+=1.0/NC/NM
for j,y in enumerate(coords):
print(data[y.id]["coord"])
print(data[y.id]["values"])
'''
def compatible_with_group(i,group):
for j in group:
if not compatible(i,j):
return False
return True
