def main(inputfile, outputfile, gfffile, gffmin, gffmax, takeStop, upstream,
downstream, verbose):
takeStart = True
if takeStop:
takeStart = False
sites = ParclipSiteContainer()
sites.loadFromFile(inputfile)
anno = gff.GFF(gfffile)
anno.filterSize(gffmin, gffmax)
anno.getChromosomePositions()
if anno.size() < 10:
print('Warning: Low number of annotation enries! ' + str(anno.size()))
fsites = ParclipSiteContainer()
percent_old = 0
percent_new = 0
for i in range(sites.size()):
if anno.isAround(sites.chrs[i], sites.pos[i], sites.strand[i],
takeStart, upstream, downstream)[1]:
fsites.addSite(sites.chrs[i], sites.pos[i], sites.m[i], sites.r[i],
sites.result[i], sites.strand[i], sites.occ[i])
percent_new = round(i / sites.size() * 100)
if percent_new > percent_old:
if verbose:
functions.showProgress(i, anno.size(), 'selecting sites')
percent_old = percent_new
fsites.save2File(outputfile)
def main(inputfile, outputfile):
if os.path.isfile(inputfile) == False:
print('Inputfile: '+inputfile+' does not exist')
sys.exit(-1)
sites = ParclipSiteContainer()
sites.loadFromFile(inputfile)
for i in range(sites.size()):
sites.occ[i] = sites.m[i]/sites.r[i]
sites.save2File(outputfile)
def main(input_file, output_file, q):
if not 0 <= q < 1:
print('q must lie between 0 and 1 - got %s' % q)
sys.exit(1)
sites = ParclipSiteContainer.from_file(input_file)
# dirty hack to avoid errors on empty files
occ_vals = []
for rec in sites:
occ_vals.append(rec.occupancy)
if len(occ_vals) > 0:
max_occ = functions.getQuantile(occ_vals, q)
records = []
for rec in sites:
if rec.occupancy > max_occ:
rec = rec._replace(occupancy=max_occ)
records.append(rec)
new_sites = ParclipSiteContainer(records)
new_sites.save2File(output_file)