def main(parclipfile, outputfile, gfffile, downstream, upstream, gene, sense, minSize,
maxSize, verbose, vstring=''):
anno = gff.GFF(gfffile)
anno.filterSize(minSize, maxSize)
pc = ParclipSiteContainer()
pc.loadFromFile(parclipfile)
with open(outputfile, 'w') as fc_out:
for g in range(anno.size()):
if verbose:
functions.showProgress(g, (anno.size() - 1), vstring)
if anno.strand[g] == '+':
values_upstream = pc.getValues(anno.chr[g], anno.start[g], anno.strand[g],
sense, upstream, gene)
values_dostream = pc.getValues(anno.chr[g], anno.stop[g], anno.strand[g],
sense, gene, downstream)
else:
values_upstream = pc.getValues(anno.chr[g], anno.stop[g], anno.strand[g],
sense, upstream, gene)
values_dostream = pc.getValues(anno.chr[g], anno.start[g], anno.strand[g],
sense, gene, downstream)
if values_upstream is not None and values_dostream is not None:
print(*chain(values_upstream, values_dostream), sep='\t', file=fc_out)
if verbose:
print()
def main(parclipA, parclipB, start, stop, width, anno=None, annowidth=100,
logRatio=False, verbose=False):
tmpA = ParclipSiteContainer()
dataB = ParclipSiteContainer()
tmpA.loadFromFile(parclipA)
tmpA.sort(key='occ')
dataB.loadFromFile(parclipB)
if start < 0 or stop < start or stop >= tmpA.size():
print('Bullshit start and stop indices. Come on! Concentrate!')
sys.exit()
dataA = parclipsites.ParclipSites('')
total = stop - start
count = 0
i = start
while count < total and i < (tmpA.size()-1):
if verbose:
functions.showProgress(count,total-1,'Selecting PAR-CLIP sites')
if anno == None:
dataA.addSite(tmpA.chrs[i], tmpA.pos[i], tmpA.m[i], tmpA.r[i],
tmpA.result[i], tmpA.strand[i], tmpA.occ[i])
count +=1
else:
if anno.isInside(tmpA.chrs[i], tmpA.pos[i], tmpA.strand[i],
annowidth, annowidth)[1]:
dataA.addSite(tmpA.chrs[i], tmpA.pos[i], tmpA.m[i], tmpA.r[i],
tmpA.result[i], tmpA.strand[i], tmpA.occ[i])
count +=1
i += 1
coloc = 1
count_coloc = 1
if verbose:
print('\n')
for i in range(dataA.size()):
values = dataB.getValues(dataA.chrs[i], dataA.pos[i], dataA.strand[i], True, width, width)
if values != None:
count_coloc += 1
coloc += max(values)
if verbose:
functions.showProgress(i, (dataA.size()-1), 'Collecting colocolization data')
coloc = coloc / count_coloc
if verbose:
print('')
if logRatio:
return math.log( coloc/functions.getQuantile(dataB.occ,0.5) ,2)
else:
return coloc