def get_motif_scores(fa, motifs):
s = Scanner()
s.set_motifs(motifs)
s.set_threshold(threshold=0.0)
seqs = Fasta(fa.seqfn)
for i, result in enumerate(s.scan(seqs, nreport=1)):
intron_id = seqs.ids[i]
for m, matches in enumerate(result):
motif = motifs[m]
for score, pos, strand in matches:
if score < 0:
score_rescaled = rescale(score,
orig_range=[motif.min_score, 0],
new_range=[0, 50])
else:
score_rescaled = rescale(score,
orig_range=[0, motif.max_score],
new_range=[50, 100])
yield (intron_id, motif.id, score_rescaled)
def command_scan(inputfile, pwmfile, nreport=1, cutoff=0.9, bed=False,
scan_rc=True, table=False, score_table=False, moods=False,
pvalue=None, bgfile=None, genome=None):
motifs = pwmfile_to_motifs(pwmfile)
index_dir = None
if genome is not None:
index_dir = os.path.join(MotifConfig().get_index_dir(), genome)
# initialize scanner
s = Scanner()
s.set_motifs(pwmfile)
fa = as_fasta(inputfile, index_dir)
if moods:
result_it = scan_it_moods(inputfile, motifs, cutoff, bgfile, nreport, scan_rc, pvalue, table)
else:
result_it = s.scan(fa, nreport, scan_rc, cutoff)
if table:
# header
yield "\t{}".format("\t".join([m.id for m in motifs]))
if moods:
result_it = scan_it_moods(inputfile, motifs, cutoff, bgfile, nreport, scan_rc, pvalue, table)
for seq_id, counts in result_it:
yield "{}\t{}".format(seq_id, "\t".join([str(x) for x in counts]))
else:
# get iterator
result_it = s.count(fa, nreport, scan_rc, cutoff)
# counts table
for i, counts in enumerate(result_it):
yield "{}\t{}".format(
fa.ids[i],
"\t".join([str(x) for x in counts])
)
elif score_table:
# get iterator
result_it = s.best_score(fa, scan_rc)
# header
yield "\t{}".format("\t".join([m.id for m in motifs]))
# score table
for i,scores in enumerate(result_it):
yield "{}\t{}".format(
fa.ids[i],
"\t".join([str(x) for x in scores])
)
else:
if moods:
for motif, d in result_it:
for seq_id,matches in d.items():
for pos,score,strand in matches:
yield format_line(fa, seq_id, motif,
score, pos, strand, bed=bed)
else:
for i, result in enumerate(result_it):
seq_id = fa.ids[i]
for motif, matches in zip(motifs, result):
for (score, pos, strand) in matches:
yield format_line(fa, seq_id, motif,
score, pos, strand, bed=bed)
