def _get_sam_groups(ignored_ids_writer=None):
"""Returns grouped sam records read from SORTED_GMAP_SAM and READS_DS."""
query_len_dict = ContigSetReaderWrapper.name_to_len_dict(READS_DS)
groups = [g for g in iter_gmap_sam(sam_filename=SORTED_GMAP_SAM,
query_len_dict=query_len_dict,
min_aln_coverage=0.99, min_aln_identity=0.85,
ignored_ids_writer=ignored_ids_writer)]
return groups
def _get_sam_groups(ignored_ids_writer=None):
"""Returns grouped sam records read from SORTED_GMAP_SAM and READS_DS."""
query_len_dict = ContigSetReaderWrapper.name_to_len_dict(READS_DS)
groups = [
g for g in iter_gmap_sam(sam_filename=SORTED_GMAP_SAM,
query_len_dict=query_len_dict,
min_aln_coverage=0.99,
min_aln_identity=0.85,
ignored_ids_writer=ignored_ids_writer)
]
return groups
def run(self,
allow_extra_5exon,
skip_5_exon_alt,
ignored_ids_fn,
good_gff_fn,
bad_gff_fn,
group_fn,
tolerate_end=100):
"""
Process the whole SAM file:
(1) Group SAM records based on where they mapped to and strands
(2) Collapse records, write collapsed isoforms to *_gff_writer,
write supportive records associated with each collapsed isoforms
to group_writer.
"""
ignored_ids_writer = open(ignored_ids_fn,
'w') if ignored_ids_fn else None
good_gff_writer = CollapseGffWriter(
good_gff_fn) if good_gff_fn else None
bad_gff_writer = CollapseGffWriter(bad_gff_fn) if bad_gff_fn else None
group_writer = GroupWriter(group_fn) if group_fn else None
cuff_index = 1
for recs in iter_gmap_sam(sam_filename=self.sam_filename,
query_len_dict=self.isoform_len_dict,
min_aln_coverage=self.min_aln_coverage,
min_aln_identity=self.min_aln_identity,
ignored_ids_writer=ignored_ids_writer):
# Iterate over groups of overlapping SAM records
for records in recs.itervalues():
if len(records) > 0:
# records: a list of overlapping SAM records, same strands
collapse_sam_records(records=records,
cuff_index=cuff_index,
cov_threshold=self.cov_threshold,
allow_extra_5exon=allow_extra_5exon,
skip_5_exon_alt=skip_5_exon_alt,
good_gff_writer=good_gff_writer,
bad_gff_writer=bad_gff_writer,
group_writer=group_writer,
tolerate_end=tolerate_end)
cuff_index += 1
# close writers.
for writer in (ignored_ids_writer, good_gff_writer, bad_gff_writer,
bad_gff_writer):
if writer:
writer.close()
def run(self, allow_extra_5exon, skip_5_exon_alt,
ignored_ids_fn, good_gff_fn, bad_gff_fn, group_fn,
tolerate_end=100):
"""
Process the whole SAM file:
(1) Group SAM records based on where they mapped to and strands
(2) Collapse records, write collapsed isoforms to *_gff_writer,
write supportive records associated with each collapsed isoforms
to group_writer.
"""
ignored_ids_writer = open(ignored_ids_fn, 'w') if ignored_ids_fn else None
good_gff_writer = CollapseGffWriter(good_gff_fn) if good_gff_fn else None
bad_gff_writer = CollapseGffWriter(bad_gff_fn) if bad_gff_fn else None
group_writer = GroupWriter(group_fn) if group_fn else None
cuff_index = 1
for recs in iter_gmap_sam(sam_filename=self.sam_filename,
query_len_dict=self.isoform_len_dict,
min_aln_coverage=self.min_aln_coverage,
min_aln_identity=self.min_aln_identity,
ignored_ids_writer=ignored_ids_writer):
# Iterate over groups of overlapping SAM records
for records in recs.itervalues():
if len(records) > 0:
# records: a list of overlapping SAM records, same strands
collapse_sam_records(records=records, cuff_index=cuff_index,
cov_threshold=self.cov_threshold,
allow_extra_5exon=allow_extra_5exon,
skip_5_exon_alt=skip_5_exon_alt,
good_gff_writer=good_gff_writer,
bad_gff_writer=bad_gff_writer,
group_writer=group_writer,
tolerate_end=tolerate_end)
cuff_index += 1
# close writers.
for writer in (ignored_ids_writer, good_gff_writer, bad_gff_writer, bad_gff_writer):
if writer:
writer.close()