def abyssmap_rmdups(in_fa, out_fa, strand_specific=False, cleanup=False, threads=1):
ids_file = in_fa + '.dup_ids'
# run abyssmap
cmd_params = ['abyss-map', '--dup']
if strand_specific:
cmd_params.append('--SS')
#endif
if threads > 1:
cmd_params.append('--threads=%d' % threads)
#endif
cmd_params.extend([in_fa, in_fa])
cmd_params.append('> %s' % ids_file)
run_shell_cmd(' '.join(cmd_params))
cids_set = set()
with open(ids_file, 'r') as fh:
for line in fh:
line_stripped = line.strip()
if len(line_stripped) > 0:
cids_set.add(line_stripped)
#endif
#endfor
#endwith
filter_fasta(in_fa, out_fa, remove_set=cids_set)
if cleanup:
os.remove(ids_file)
def bowtie2_self_align(fasta, outputsam, threads=1, strand_specific=False, path_strip_sam_seq_qual=None, preset='--sensitive', k=2):
"""Align all fasta sequences to each other with Bowtie2.
"""
# Build index files for the concatenated fasta
bt2_index_cmd_params = ['bowtie2-build --quiet', fasta, fasta]
run_shell_cmd(' '.join(bt2_index_cmd_params))
# Self-align concatenated fasta with Bowtie2
bt2_align_cmd_params = ['set -euo pipefail && bowtie2']
if strand_specific:
bt2_align_cmd_params.append('--norc')
#endif
bt2_align_cmd_params.extend([preset, '-k %d' % k, '--omit-sec-seq --end-to-end -f', '-p %d' % threads, fasta, fasta])
if path_strip_sam_seq_qual:
bt2_align_cmd_params.append('|' + path_strip_sam_seq_qual)
#endif
bt2_align_cmd_params.append('|gzip -c >' + outputsam)
run_shell_cmd(' '.join(bt2_align_cmd_params))
def blat_merge_fastas(path_prefix_map, merged_fa, concat_fa=None, concat_fa_selfalign_psl=None, percent_identity=0.95, strand_specific=False, cleanup=False, minoverlap=0, threads=1, indel_size_tolerance=1, min_seq_len=32):
"""Merge fasta files into a single fasta file by removing redundant sequences.
"""
if concat_fa is None:
concat_fa = merged_fa + '.tmp.concat.fa'
#endif
if concat_fa_selfalign_psl is None:
concat_fa_selfalign_psl = merged_fa + '.tmp.concat.psl'
#endif
# Concatenate the fastas together and give the contigs of each set a prefix
concat_fastas(path_prefix_map, concat_fa)
# Self-align concatenated fasta with Bowtie2
blat_self_align(concat_fa, concat_fa_selfalign_psl, percent_id=percent_identity, max_consecutive_edits=indel_size_tolerance, min_seq_len=min_seq_len, threads=threads)
# Identify NON-redundant contigs
nrrefs = psl_cid_extractor.extract_cids(psl=concat_fa_selfalign_psl, samestrand=strand_specific, min_percent_identity=percent_identity, max_consecutive_edits=indel_size_tolerance, report_redundant=False)
tmpfiles = []
nr_fa_long = merged_fa + '.tmp.long.fa'
nr_fa_short = None
if minoverlap > 0:
nr_fa_short = merged_fa + '.tmp.short.fa'
tmpfiles.append(nr_fa_short)
#endif
# Gather the non-contained sequences and split into 2 partitions:
# 1. shorter than (min overlap + 1)
# 2. longer than or equal to (min overlap + 1)
filter_fasta(concat_fa, nr_fa_long, min_length=minoverlap+1, keep_set=nrrefs, fasta_out_st=nr_fa_short)
# overlap-layout the long sequences
if minoverlap > 0:
# generate the sequence overlap graph
overlap_dot = merged_fa + '.tmp.long.dot'
overlap_cmd_params = ['abyss-overlap', '--threads=%d' % threads, '--min=%d' % minoverlap]
if strand_specific:
overlap_cmd_params.append('--SS')
#endif
overlap_cmd_params.append(nr_fa_long)
overlap_cmd_params.append('>' + overlap_dot)
run_shell_cmd(' '.join(overlap_cmd_params))
# layout contigs using the overlap graph
layout_path = merged_fa + '.tmp.long.path'
layout_cmd_params = ['abyss-layout', '--kmer=%d' % (minoverlap+1), '--out=%s' % layout_path]
if strand_specific:
layout_cmd_params.append('--SS')
#endif
layout_cmd_params.append(overlap_dot)
run_shell_cmd(' '.join(layout_cmd_params))
# generate fasta for O-L
overlap_fa = merged_fa + '.incomplete'
mergecontigs_cmd_params = ['MergeContigs', '--kmer=%d' % (minoverlap+1), '--out=%s' % overlap_fa, nr_fa_long, overlap_dot, layout_path]
run_shell_cmd(' '.join(mergecontigs_cmd_params))
# append the short sequences to the same fasta
with open(overlap_fa, 'a') as fout:
with open(nr_fa_short, 'r') as fin:
for line in fin:
fout.write(line)
#endfor
#endwith
#endwith
shutil.move(overlap_fa, merged_fa)
tmpfiles.extend([concat_fa, concat_fa_selfalign_psl, nr_fa_long, nr_fa_short, overlap_dot, layout_path])
else:
shutil.move(nr_fa_long, merged_fa)
tmpfiles.extend([concat_fa, concat_fa_selfalign_psl])
#endif
if cleanup and tmpfiles is not None:
for t in tmpfiles:
if t is not None and os.path.isfile(t):
os.remove(t)