def last_part():
the_donors = msgpack.load(file('donors_{}'.format(c.DATASET),'rb'))
#clean up
the_donors = [donors for donors in the_donors if donors]
good_changes = []
for donor_tuple in the_donors:
donor = donor_tuple[1]
pos = donor_tuple[0]
ref_piece = ref[pos:pos+len(donor)]
changes, score = dependencies.identify_changes(ref_piece,donor,0)
if score < 10:
#print visualize_lines(donor,pos,ref_piece)
for cc in changes:
try:
cc[3] += pos
except:
cc[2] += pos
good_changes.extend(changes)
utils.write_indels(good_changes, DO_NUMBER)
#
good_changes = []
for donor_tuple in the_donors:
donor = donor_tuple[1]
pos = donor_tuple[0]
ref_piece = ref[pos:pos+len(donor)]
changes, score = dependencies.identify_changes(ref_piece,donor,-1)
if score < len(donor) * 0.4: # thanks leah
#visualize_lines(donor,pos,ref, pos)
#print 'C.\n{}'.format(changes)
for cc in changes:
try:
if cc[2] == '.':
changes.remove(cc)
continue
cc[3] += pos
except:
cc[2] += pos
good_changes.extend(changes)
for _ in good_changes:
print _
utils.write_indels(good_changes, DO_NUMBER)
msgpack.dump(the_donors,file('donors_{}'.format(c.DATASET),'wb'))
print 'done?'
the_donors = pileup.generate_donor_pieces(sub, ref, d)
the_donors = [donors for donors in the_donors if donors]
print "done assembling, SW time..."
good_changes = []
count = 0.0
for donor_tuple in the_donors:
donor = donor_tuple[1]
pos = donor_tuple[0]
ref_piece = ref[pos : pos + len(donor)]
changes, score = dependencies.identify_changes(ref_piece, donor, -1)
if score < len(donor) * 0.4: # thanks leah
for cc in changes:
try:
if cc[2] == ".":
changes.remove(cc)
continue
cc[3] += pos
except:
cc[2] += pos
good_changes.extend(changes)
count += 1
if count % 100 == 0:
print "SW progress: {}".format(count / len(the_donors))
utils.write_indels(good_changes, job_number)
msgpack.dump(the_donors, file("donors_{}_part_{}".format(c.DATASET, job_number), "wb"))
