def add_batch(batch_index, pCS, orphans, fasta_d, cpus, dun_use_partial):
"""
1. align batch<i>.fasta against seed<i>.S.fasta, process -> write remains to batch<i>.remains.fasta
2. align batch<i>.remains.fasta against seed<i>.orphans.fasta -> write remains to batch<i>.remains2.fasta
3. self align batch<i>.remains2.fasta -> combine remains+orphans to new orphans
4. write out seed<i+1>.S.fasta and seed<i+1>.orphans.fasta
"""
cur_file = "batch{0}.fasta".format(batch_index)
seqids = set([r.id for r in SeqIO.parse(open(cur_file), 'fasta')])
o = ar.run_minimap(cur_file,
"seed{0}.S.fasta".format(batch_index),
cpus=cpus)
print("processing", o, file=sys.stderr)
pCS, remains = sp.process_align_to_pCS(o,
seqids,
pCS,
MiniReader,
dun_use_partial=dun_use_partial)
print("pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.values()), len(orphans), len(remains)), file=sys.stderr)
# write batch<i>.remains.fasta
cur_file = "batch{0}.remains.fasta".format(batch_index)
FileIO.write_seqids_to_fasta(remains, cur_file, fasta_d)
o = ar.run_minimap(cur_file,
"seed{0}.orphans.fasta".format(batch_index),
cpus=cpus)
print("processing", o, file=sys.stderr)
pCS, orphans, remains = sp.process_align_to_orphan(
o, remains, orphans, pCS, MiniReader, dun_use_partial=dun_use_partial)
print("pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.values()), len(orphans), len(remains)), file=sys.stderr)
# write batch<i>.remains2.fasta and self align
cur_file = "batch{0}.remains2.fasta".format(batch_index)
FileIO.write_seqids_to_fasta(remains, cur_file, fasta_d)
o = ar.run_minimap(cur_file, cur_file, cpus=cpus)
print("processing", o, file=sys.stderr)
pCS, remains = sp.process_self_align_into_seed(
o, remains, MiniReader, pCS, dun_use_partial=dun_use_partial)
print("pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.values()), len(orphans), len(remains)), file=sys.stderr)
# combine remains+orphans to new orphans
orphans = orphans.union(remains)
FileIO.write_preClusterSet_to_fasta(
pCS, "seed{0}.S.fasta".format(batch_index + 1), fasta_d)
FileIO.write_seqids_to_fasta(
orphans, "seed{0}.orphans.fasta".format(batch_index + 1), fasta_d)
return pCS, orphans
def add_batch(batch_index, pCS, orphans, fasta_d, cpus, dun_use_partial):
"""
1. align batch<i>.fasta against seed<i>.S.fasta, process -> write remains to batch<i>.remains.fasta
2. align batch<i>.remains.fasta against seed<i>.orphans.fasta -> write remains to batch<i>.remains2.fasta
3. self align batch<i>.remains2.fasta -> combine remains+orphans to new orphans
4. write out seed<i+1>.S.fasta and seed<i+1>.orphans.fasta
"""
cur_file = "batch{0}.fasta".format(batch_index)
seqids = set([r.id for r in SeqIO.parse(open(cur_file), 'fasta')])
o = ar.run_minimap(cur_file, "seed{0}.S.fasta".format(batch_index), cpus=cpus)
print >> sys.stderr, "processing", o
pCS, remains = sp.process_align_to_pCS(o, seqids, pCS, MiniReader, dun_use_partial=dun_use_partial)
print >> sys.stderr, "pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.itervalues()), len(orphans), len(remains))
# write batch<i>.remains.fasta
cur_file = "batch{0}.remains.fasta".format(batch_index)
FileIO.write_seqids_to_fasta(remains, cur_file, fasta_d)
o = ar.run_minimap(cur_file, "seed{0}.orphans.fasta".format(batch_index), cpus=cpus)
print >> sys.stderr, "processing", o
pCS, orphans, remains = sp.process_align_to_orphan(o, remains, orphans, pCS, MiniReader, dun_use_partial=dun_use_partial)
print >> sys.stderr, "pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.itervalues()), len(orphans), len(remains))
# write batch<i>.remains2.fasta and self align
cur_file = "batch{0}.remains2.fasta".format(batch_index)
FileIO.write_seqids_to_fasta(remains, cur_file, fasta_d)
o = ar.run_minimap(cur_file, cur_file, cpus=cpus)
print >> sys.stderr, "processing", o
pCS, remains = sp.process_self_align_into_seed(o, remains, MiniReader, pCS, dun_use_partial=dun_use_partial)
print >> sys.stderr, "pCS: {0}, tucked: {1}, orphans: {2}, remains: {3}".format( \
len(pCS.S), sum(v == 'T' for v in pCS.seq_stat.itervalues()), len(orphans), len(remains))
# combine remains+orphans to new orphans
orphans = orphans.union(remains)
FileIO.write_preClusterSet_to_fasta(pCS, "seed{0}.S.fasta".format(batch_index+1), fasta_d)
FileIO.write_seqids_to_fasta(orphans, "seed{0}.orphans.fasta".format(batch_index+1), fasta_d)
return pCS, orphans