def do_buffer(gpd_lines,fasta,args):
results = []
for gpd_line in gpd_lines:
gpd = GPD(gpd_line)
l = gpd.get_length()
if l < args.length: continue
num = int(float(l)/float(args.length))
rem = l % args.length
#print 'rem : '+str(rem)
extra = 0
offset = 0
#if space > 1: # we have room to make multiple passes
# #print '---'
# #print 'length: '+str(l)
# #print 'strand: '+gpd.get_strand()
# if random.random() < 0.5: extra = rem
# offset = int(float(args.length)/float(args.coverage))
#else:
# offset = int(float(rem)/float(args.coverage))
if args.short_reads:
offset = 0
if random.random() < 0.5: offset = rem
gsub = gpd.subset(offset,args.length+offset)
#print gsub.get_gpd_line()
val = get_sam(gsub,fasta)
results.append(val)
#continue
else:# not short reads
for i in range(0,args.coverage):
init = 0
if num == 0 and rem > 0:
init = random.choice(range(0,rem))
elif num > 0:
init = random.choice(range(0,args.length))
#start = (i*offset+extra) % args.length
#while start+args.length <= l:
for j in range(init,l,args.length):
if j + args.length > l: break
#print str(start)+" "+str(start+args.length)
gsub = gpd.subset(j,j+args.length)
val = get_sam(gsub,fasta)
results.append(val)
#print gsub.get_sequence(fasta)
#start += args.length
#print gsub.get_strand()
#print space
#print rem
#print gpd
return results
def do_buffer(gpd_lines, fasta, args):
results = []
for gpd_line in gpd_lines:
gpd = GPD(gpd_line)
l = gpd.get_length()
if l < args.length: continue
num = int(float(l) / float(args.length))
rem = l % args.length
#print 'rem : '+str(rem)
extra = 0
offset = 0
#if space > 1: # we have room to make multiple passes
# #print '---'
# #print 'length: '+str(l)
# #print 'strand: '+gpd.get_strand()
# if random.random() < 0.5: extra = rem
# offset = int(float(args.length)/float(args.coverage))
#else:
# offset = int(float(rem)/float(args.coverage))
if args.short_reads:
offset = 0
if random.random() < 0.5: offset = rem
gsub = gpd.subset(offset, args.length + offset)
#print gsub.get_gpd_line()
val = get_sam(gsub, fasta)
results.append(val)
#continue
else: # not short reads
for i in range(0, args.coverage):
init = 0
if num == 0 and rem > 0:
init = random.choice(range(0, rem))
elif num > 0:
init = random.choice(range(0, args.length))
#start = (i*offset+extra) % args.length
#while start+args.length <= l:
for j in range(init, l, args.length):
if j + args.length > l: break
#print str(start)+" "+str(start+args.length)
gsub = gpd.subset(j, j + args.length)
val = get_sam(gsub, fasta)
results.append(val)
#print gsub.get_sequence(fasta)
#start += args.length
#print gsub.get_strand()
#print space
#print rem
#print gpd
return results
def main():
#do our inputs
args = do_inputs()
sys.stderr.write("Reading reference genepred\n")
ref = {}
tx_strand = {}
z = 0
with open(args.reference_genepred) as inf:
for line in inf:
gpd = GPD(line)
gname = gpd.get_gene_name()
tname = gpd.get_transcript_name()
tx_strand[tname] = gpd.get_strand()
if gname not in ref: ref[gname] = []
ref[gname].append(gpd)
z += 1
sys.stderr.write("Read "+str(len(ref.keys()))+" genes and "+str(z)+" transcripts\n")
if args.maximum_isoforms > 0:
sys.stderr.write("Removing genes with more than "+str(args.maximum_isoforms)+" isoforms.\n")
for gname in ref.keys():
if len(ref[gname]) > args.maximum_isoforms: del ref[gname]
sys.stderr.write("Now have "+str(len(ref.keys()))+" genes and "+str(sum([len(ref[x]) for x in ref.keys()]))+" transcripts\n")
sys.stderr.write("Filtering by length "+str(args.minimum_length)+" bp\n")
for gname in ref.keys():
passing = []
for gpd in ref[gname]:
if gpd.get_length() < args.minimum_length: continue
passing.append(gpd)
if len(passing) == 0: del ref[gname]
else: ref[gname] = passing
sys.stderr.write("Now have "+str(len(ref.keys()))+" genes and "+str(sum([len(ref[x]) for x in ref.keys()]))+" transcripts\n")
sys.stderr.write("Converting gpd into exon bed\n")
beds = []
for gname in ref.keys():
for gpd in ref[gname]:
tname = gpd.get_transcript_name()
for i in range(0,len(gpd.exons)):
ex = gpd.exons[i]
beds.append(ex.get_range().get_bed_array()+[gname,tname,i])
with open(args.tempdir+'/gpd.bed','w') as of:
for bed in sorted(beds,key=lambda x: (x[0],x[1],x[2],x[3],x[4],x[5])):
of.write("\t".join([str(x) for x in bed])+"\n")
sys.stderr.write("intersecting with bed depth\n")
of = open(args.tempdir+'/intersect.bed','w')
cmd = 'bedtools intersect -wo -a - -b '+args.tempdir+'/gpd.bed'
p = Popen(cmd.split(),stdin=args.bed_depth,stdout=of)
p.communicate()
coverage = {}
sys.stderr.write("Reading the intersection\n")
with open(args.tempdir+'/intersect.bed') as inf:
for line in inf:
f = line.rstrip().split("\t")
gname = f[7]
tname = f[8]
depth = int(f[3])
bed1 = Bed(f[0],int(f[1]),int(f[2]))
bed2 = Bed(f[4],int(f[5]),int(f[6]))
bed = bed1.union(bed2)
bed.set_payload(depth)
if gname not in coverage:
coverage[gname] = {}
if tname not in coverage[gname]:
coverage[gname][tname] = []
coverage[gname][tname].append(bed)
transcript_depths = {}
for gname in coverage:
for tname in coverage[gname]:
ref_gpd = [x for x in ref[gname] if x.get_transcript_name()==tname][0]
rlen = ref_gpd.get_length()
bases_covered = sum([x.length() for x in coverage[gname][tname]])
bases_area = sum([x.length()*x.get_payload() for x in coverage[gname][tname]])
avg_depth = float(bases_area)/float(rlen)
if avg_depth < args.minimum_average_depth: continue
if bases_covered < args.minimum_length: continue
#print gname
#print tname
#print rlen
#print bases_covered
#print bases_area
total_positions = {}
for ex in ref_gpd.exons:
b = ex.get_range().get_bed_array()
for i in range(b[1],b[2]):
total_positions[i] = 0 # zero indexed
for b in coverage[gname][tname]:
depth = b.get_payload()
barr = b.get_bed_array()
for i in range(barr[1],barr[2]):
total_positions[i] = depth
transcript_depths[tname] = total_positions
sys.stderr.write("have information needed to plot from "+str(len(transcript_depths.keys()))+" transcripts\n")
outputs = []
for tname in transcript_depths:
depths = transcript_depths[tname]
positions = sorted(depths.keys())
tx_len = len(positions)
bins = {}
for i in range(0,tx_len):
bin = int(100*float(i)/float(tx_len))
if bin not in bins: bins[bin] = []
bins[bin].append(depths[positions[i]])
for bin in bins:
bins[bin] = average(bins[bin])
biggest = float(max(bins.values()))
tx_array = [float(bins[x])/biggest for x in sorted(bins.keys())]
if tx_strand[tname] == '-':
tx_array.reverse()
#outputs.append(tx_array)
args.output.write(tname+"\t"+"\t".join([str(x) for x in tx_array])+"\n")
#for i in range(0,100):
# args.output.write("\t".join([str(x[i]) for x in outputs])+"\n")
args.output.close()
# Temporary working directory step 3 of 3 - Cleanup
if not args.specific_tempdir:
rmtree(args.tempdir)
def annotate_line(inputs):
global txome
(line, z, args) = inputs
gpd = GPD(line)
gpd.set_payload(z)
v = gpd.get_range()
if v.chr not in txome: return None
possible = [x.get_payload() for x in txome[v.chr] if x.overlaps(v)]
candidates = []
if len(possible) == 0: return None
for tx in possible:
eo = None
full = False
subset = False
econsec = 1
if tx.get_exon_count() == 1 or gpd.get_exon_count() == 1:
eo = gpd.exon_overlap(tx, single_minover=100, single_frac=0.5)
else:
eo = gpd.exon_overlap(tx,
multi_minover=10,
multi_endfrac=0,
multi_midfrac=0.8,
multi_consec=False)
if eo.is_full_overlap():
full = True
if eo.is_subset():
subset = True
if eo:
econsec = eo.consecutive_exon_count()
if not eo: continue
ecnt = eo.match_exon_count()
osize = gpd.overlap_size(tx)
candidates.append([
full, subset, ecnt, econsec,
gpd.get_exon_count(),
tx.get_exon_count(), osize,
gpd.get_length(),
tx.get_length(), tx
])
if len(candidates) == 0: return None
bests = sorted(candidates,
key=lambda x: (-x[0], -x[1], -x[3], -x[2], -min(
float(x[6]) / float(x[7]),
float(x[6]) / float(x[8]))))
#line_z
v = bests[0]
### we have the annotation
z = gpd.get_payload()
#line = line_z[0]
#gpd = GPD(line)
if not v: return None
type = 'partial'
if v[0]: type = 'full'
exon_count = v[2]
most_consecutive_exons = v[3]
read_exon_count = v[4]
tx_exon_count = v[5]
overlap_size = v[6]
read_length = v[7]
tx_length = v[8]
return str(z)+"\t"+gpd.get_transcript_name()+"\t"+v[9].get_gene_name()+"\t"+v[9].get_transcript_name()+"\t"+type+"\t"+\
str(exon_count)+"\t"+str(most_consecutive_exons)+"\t"+str(read_exon_count)+"\t"+str(tx_exon_count)+"\t"+\
str(overlap_size)+"\t"+str(read_length)+"\t"+str(tx_length)+"\t"+gpd.get_range().get_range_string()+"\t"+v[9].get_range().get_range_string()+"\t"+str(v[9].get_payload())+"\n"
def annotate_line(inputs):
global txome
(line,z,args) = inputs
gpd = GPD(line)
gpd.set_payload(z)
v = gpd.get_range()
if v.chr not in txome: return None
possible = [x.get_payload() for x in txome[v.chr] if x.overlaps(v)]
candidates = []
if len(possible) == 0: return None
for tx in possible:
eo = None
full = False
subset = False
econsec = 1
if tx.get_exon_count() == 1 or gpd.get_exon_count() == 1:
eo = gpd.exon_overlap(tx,single_minover=100,single_frac=0.5)
else:
eo = gpd.exon_overlap(tx,multi_minover=10,multi_endfrac=0,multi_midfrac=0.8,multi_consec=False)
if eo.is_full_overlap():
full = True
if eo.is_subset():
subset = True
if eo:
econsec = eo.consecutive_exon_count()
if not eo: continue
ecnt = eo.match_exon_count()
osize = gpd.overlap_size(tx)
candidates.append([full,subset,ecnt,econsec,gpd.get_exon_count(),tx.get_exon_count(),osize,gpd.get_length(),tx.get_length(),tx])
if len(candidates)==0: return None
bests = sorted(candidates,key=lambda x: (-x[0],-x[1],-x[3],-x[2],-min(float(x[6])/float(x[7]),float(x[6])/float(x[8]))))
#line_z
v = bests[0]
### we have the annotation
z = gpd.get_payload()
#line = line_z[0]
#gpd = GPD(line)
if not v: return None
type = 'partial'
if v[0]: type = 'full'
exon_count = v[2]
most_consecutive_exons = v[3]
read_exon_count = v[4]
tx_exon_count = v[5]
overlap_size = v[6]
read_length = v[7]
tx_length = v[8]
return str(z)+"\t"+gpd.get_transcript_name()+"\t"+v[9].get_gene_name()+"\t"+v[9].get_transcript_name()+"\t"+type+"\t"+\
str(exon_count)+"\t"+str(most_consecutive_exons)+"\t"+str(read_exon_count)+"\t"+str(tx_exon_count)+"\t"+\
str(overlap_size)+"\t"+str(read_length)+"\t"+str(tx_length)+"\t"+gpd.get_range().get_range_string()+"\t"+v[9].get_range().get_range_string()+"\t"+str(v[9].get_payload())+"\n"
