def count_flank(beds,bp,strand,upstream=True):
'''
flank seq aggregation ( no splicing)
if upstream is False , count downstream
'''
pos_sum=[0 for i in xrange(bp)]
pos_dis=[[] for i in xrange(bp)]
neg_sum=[0 for i in xrange(bp)]
neg_dis=[[] for i in xrange(bp)]
for bed in beds:
if upstream:
flank=bed.upstream(bp)
offset=bp-flank.cdna_length() # in case upstream is less than bp
else:
flank=bed.downstream(bp)
offset=0
pos=[0 for i in xrange(bp)]
neg=[0 for i in xrange(bp)]
for read in bam.query(flank,"bam1",strand=strand):
translated_read=translate_coordinates(flank,read)
if translated_read.strand=="+" or translated_read.strand==".":
for start,size in itertools.izip(translated_read.blockStarts,translated_read.blockSizes):
for j in xrange(start+offset,start+size+offset):
if j>=0 and j<bp:
pos[j]+=1
else:
for start,size in itertools.izip(translated_read.blockStarts,translated_read.blockSizes):
for j in xrange(start+offset,start+size+offset):
if j>=0 and j<bp:
neg[j]+=1
for i in xrange(bp):
pos_sum[i]+=pos[i]
neg_sum[i]+=neg[i]
pos_dis[i].append(pos[i])
neg_dis[i].append(neg[i])
return pos_sum,pos_dis,neg_sum,neg_dis
def run(args):
logging.basicConfig(level=logging.INFO)
up=args.up
down=args.down
bp_num=up+down
offset=-up
bam=DBI.init(args.bam,"bam")
fin=IO.fopen(args.input,"r")
out=IO.fopen(args.output,"w")
bin_sum=[0 for i in xrange(bp_num)]
bin_e=[0.0 for i in xrange(bp_num)]
bin_dis=[[] for i in xrange(bp_num)]
for i0,bed in enumerate(TableIO.parse(fin,args.format)):
bed_bin=[0 for i in xrange(bp_num)]
if args.tts:
pos=bed.tts()
else:
pos=bed.tss()
pos_flank=get_flank_region(pos,up,down)
for read in bam.query(pos_flank,"bam1",strand="read1"):
a=translate_coordinates(pos,read)
#print(a,file=out)
for e in a.Exons():
#print(e,file=out)
start=e.start-offset
end=e.stop-offset
if start < 0: start=0
if end > bp_num: end=bp_num
for j in xrange(start,end):
bed_bin[j]+=1
for i in xrange(bp_num):
bin_sum[i]+=bed_bin[i]
bin_dis[i].append(bed_bin[i])
bed_num=i0+1
for i in xrange(bp_num):
bin_e[i]=gini_coefficient(bin_dis[i])
if args.tts:
print("pos_to_tts\taggregation_mean\tgini_coefficient",file=out)
else:
print("pos_to_tss\taggregation_mean\tgini_coefficient",file=out)
for i in xrange(bp_num):
print("{bin}\t{aggregation}\t{E}".format(bin=i+offset,aggregation=float(bin_sum[i])/bed_num,E=bin_e[i]),file=out)
try:
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
matplotlib.rcParams.update({'font.size':9})
ax1=plt.subplot2grid((7,1),(6,0))
plt.ylabel('gini coeffecient')
plt.fill_between(range(-up,down),bin_e,color="r",alpha=0.2,y2=0)
ax1.set_ylim(0,1)
ax1.set_xlim(-up,down)
ax1.axes.get_xaxis().set_visible(False)
plt.axvline(x=0,linewidth=1, color='y')
ax2=plt.subplot2grid((7,1),(0,0),rowspan=5)
ax2.set_xlim(-up,down)
plt.plot(range(-up,down),[float(i)/bed_num for i in bin_sum])
plt.ylabel('mean coverage')
if args.tts:
plt.xlabel('pos to tts (bp)')
else:
plt.xlabel('pos to tss (bp)')
plt.axvline(x=0,linewidth=1, color='y')
plt.grid(True)
plt.savefig(args.output+".png")
except:
pass