def check_already_decontaminated(sample, fragment, PCR):
"""Check from the summary and output file whether decontamination has been done"""
import os
from hivwholeseq.utils.mapping import get_number_reads
if (fragment == "F4") and (sample.name in maj_contnames):
return True
fn_sum = get_decontaminate_summary_filename(sample.patient, sample.name, fragment, PCR=PCR)
fn_in = sample.get_mapped_filtered_filename(fragment, PCR=PCR, decontaminated=False)
fn_out = sample.get_mapped_filtered_filename(fragment, PCR=PCR, decontaminated=True)
if not all(map(os.path.isfile, [fn_sum, fn_in, fn_out])):
print sample.patient, sample.name, fragment, PCR
return False
n_reads_in = get_number_reads(fn_in)
n_reads_out = get_number_reads(fn_out)
(n_reads_good, n_reads_cont) = get_number_reads_summary(fn_sum)
if n_reads_out != n_reads_good:
print sample.patient, sample.name, fragment, PCR, n_reads_in, n_reads_out, n_reads_good, n_reads_cont
return False
if n_reads_good < 0.5 * n_reads_in:
print sample.patient, sample.name, fragment, PCR, n_reads_in, n_reads_out, n_reads_good, n_reads_cont
return False
return True
def check_already_decontaminated(sample, fragment, PCR):
'''Check from the summary and output file whether decontamination has been done'''
import os
from hivwholeseq.utils.mapping import get_number_reads
if (fragment == 'F4') and (sample.name in maj_contnames):
return True
fn_sum = get_decontaminate_summary_filename(sample.patient,
sample.name,
fragment,
PCR=PCR)
fn_in = sample.get_mapped_filtered_filename(fragment,
PCR=PCR,
decontaminated=False)
fn_out = sample.get_mapped_filtered_filename(fragment,
PCR=PCR,
decontaminated=True)
if not all(map(os.path.isfile, [fn_sum, fn_in, fn_out])):
print sample.patient, sample.name, fragment, PCR
return False
n_reads_in = get_number_reads(fn_in)
n_reads_out = get_number_reads(fn_out)
(n_reads_good, n_reads_cont) = get_number_reads_summary(fn_sum)
if n_reads_out != n_reads_good:
print sample.patient, sample.name, fragment, PCR, n_reads_in, n_reads_out, n_reads_good, n_reads_cont
return False
if n_reads_good < 0.5 * n_reads_in:
print sample.patient, sample.name, fragment, PCR, n_reads_in, n_reads_out, n_reads_good, n_reads_cont
return False
return True
if samples.loc[contname].patient == pname:
del consensi_sample[contname]
print samplename,
if VERBOSE >= 2:
print ""
(n_good, n_cont) = filter_contamination(
bamfilename, bamfilename_out, consensi_sample, samplename, VERBOSE=VERBOSE, maxreads=maxreads
)
if VERBOSE:
print "good:", n_good, "contaminated:", n_cont
if summary:
sfn = get_decontaminate_summary_filename(pname, samplename, fragment, PCR=PCR_sample)
with open(sfn, "w") as f:
f.write(
"Call: python decontaminate_reads.py"
+ " --samples "
+ samplename
+ " --fragments "
+ fragment
+ " --verbose "
+ str(VERBOSE)
)
if maxreads != -1:
f.write(" --maxreads " + str(maxreads))
f.write("\n")
f.write("Good: " + str(n_good) + "\n")
f.write("Contaminated: " + str(sum(n_cont.itervalues())) + "\n")
print samplename,
if VERBOSE >= 2:
print ''
(n_good, n_cont) = filter_contamination(bamfilename,
bamfilename_out,
consensi_sample,
samplename,
VERBOSE=VERBOSE,
maxreads=maxreads)
if VERBOSE:
print 'good:', n_good, 'contaminated:', n_cont
if summary:
sfn = get_decontaminate_summary_filename(pname,
samplename,
fragment,
PCR=PCR_sample)
with open(sfn, 'w') as f:
f.write('Call: python decontaminate_reads.py'+\
' --samples '+samplename+\
' --fragments '+fragment+\
' --verbose '+str(VERBOSE))
if maxreads != -1:
f.write(' --maxreads ' + str(maxreads))
f.write('\n')
f.write('Good: ' + str(n_good) + '\n')
f.write('Contaminated: ' +
str(sum(n_cont.itervalues())) + '\n')
f.write('Contamination sources:\n')
for contname, n_conti in n_cont.iteritems():
f.write('{:<20s}'.format(contname) + ' ' +
samples = load_samples_sequenced()
samplenames = refnames + samples.index.tolist()
for fragment in fragments:
if VERBOSE >= 1:
print fragment
mat = np.ma.masked_all((len(samplenames), len(samplenames)))
for (samplename, sample) in samples.iterrows():
if VERBOSE >= 2:
print samplename,
isp = samplenames.index(samplename)
sample = SamplePat(sample)
fn = get_decontaminate_summary_filename(sample.patient, samplename, fragment,
PCR=1)
#FIXME: still finishing this one
if samplename == '29184':
continue
if not os.path.isfile(fn):
if VERBOSE >= 2:
print 'summary file not found'
continue
n_good, n_cont, n_cont_dict = get_number_reads_summary(fn, details=True)
if VERBOSE >= 2:
print n_good, n_cont