def test_collapse_fuzzy_junctions(self):
"""Test collapse_fuzzy_junctions, can_merge and compare_fuzzy_junctions."""
test_name = "collapse_fuzzy_junctions"
input_gff = op.join(_DAT_DIR_, "input_%s.gff" % test_name)
input_group = op.join(_DAT_DIR_, "input_%s.group.txt" % test_name)
output_gff = op.join(_OUT_DIR_, "output_%s.gff" % test_name)
output_group = op.join(_OUT_DIR_, "output_%s.group.txt" % test_name)
records = [r for r in CollapseGffReader(input_gff)]
self.assertEqual(len(records), 4)
r0, r1, r2, r3 = records
# comparing r0 and r1
m = compare_fuzzy_junctions(r0.ref_exons, r1.ref_exons, max_fuzzy_junction=5)
self.assertEqual(m, "subset")
self.assertTrue(can_merge(m, r0, r1, allow_extra_5exon=True, max_fuzzy_junction=5))
# comparing r2 and r3
m = compare_fuzzy_junctions(r2.ref_exons, r3.ref_exons, max_fuzzy_junction=5)
self.assertEqual(m, "exact")
self.assertTrue(can_merge(m, r2, r3, allow_extra_5exon=True, max_fuzzy_junction=5))
# call collapse_fuzzy_junctions and write fuzzy output.
collapse_fuzzy_junctions(gff_filename=input_gff,
group_filename=input_group,
fuzzy_gff_filename=output_gff,
fuzzy_group_filename=output_group,
allow_extra_5exon=True,
max_fuzzy_junction=5)
r4, r5 = [r for r in CollapseGffReader(output_gff)]
self.assertEqual(r1, r4)
self.assertEqual(r3, r5)
def test_collapse_fuzzy_junctions(self):
"""Test collapse_fuzzy_junctions, can_merge and compare_fuzzy_junctions."""
test_name = "collapse_fuzzy_junctions"
input_gff = op.join(_DAT_DIR_, "input_%s.gff" % test_name)
input_group = op.join(_DAT_DIR_, "input_%s.group.txt" % test_name)
output_gff = op.join(_OUT_DIR_, "output_%s.gff" % test_name)
output_group = op.join(_OUT_DIR_, "output_%s.group.txt" % test_name)
records = [r for r in CollapseGffReader(input_gff)]
self.assertEqual(len(records), 4)
r0, r1, r2, r3 = records
# comparing r0 and r1
m = compare_fuzzy_junctions(r0.ref_exons, r1.ref_exons, max_fuzzy_junction=5)
self.assertEqual(m, "subset")
self.assertTrue(can_merge(m, r0, r1, allow_extra_5exon=True, max_fuzzy_junction=5))
# comparing r2 and r3
m = compare_fuzzy_junctions(r2.ref_exons, r3.ref_exons, max_fuzzy_junction=5)
self.assertEqual(m, "exact")
self.assertTrue(can_merge(m, r2, r3, allow_extra_5exon=True, max_fuzzy_junction=5))
# call collapse_fuzzy_junctions and write fuzzy output.
collapse_fuzzy_junctions(gff_filename=input_gff,
group_filename=input_group,
fuzzy_gff_filename=output_gff,
fuzzy_group_filename=output_group,
allow_extra_5exon=True,
max_fuzzy_junction=5)
r4, r5 = [r for r in CollapseGffReader(output_gff)]
self.assertEqual(r1, r4)
self.assertEqual(r3, r5)
def run(self):
"""
First, collapse input isoforms by calling Branch.run().
Then collapse fuzzy junctions by calling collapse_fuzzy_junctions.
Finally, pick up representitive gff record for each group of collapsed isoforms.
"""
self.validate_inputs()
logging.info("Collapsing isoforms into transcripts.")
b = Branch(isoform_filename=self.isoform_filename,
sam_filename=self.sam_filename,
cov_threshold=self.min_flnc_coverage,
min_aln_coverage=self.min_aln_coverage,
min_aln_identity=self.min_aln_identity)
b.run(allow_extra_5exon=self.allow_extra_5exon,
skip_5_exon_alt=self.skip_5_exon_alt,
ignored_ids_fn=self.ignored_ids_txt_fn,
good_gff_fn=self.good_unfuzzy_gff_fn,
bad_gff_fn=self.bad_unfuzzy_gff_fn,
group_fn=self.unfuzzy_group_fn)
logging.info("Good unfuzzy isoforms written to: %s", realpath(self.good_unfuzzy_gff_fn))
logging.info("Bad unfuzzy isoforms written to: %s", realpath(self.bad_unfuzzy_gff_fn))
logging.info("Unfuzzy isoform groups written to: %s", realpath(self.unfuzzy_group_fn))
if self.shall_collapse_fuzzy_junctions:
logging.info("Further collapsing fuzzy junctions.")
# need to further collapse those that have fuzzy junctions!
collapse_fuzzy_junctions(gff_filename=self.good_unfuzzy_gff_fn,
group_filename=self.unfuzzy_group_fn,
fuzzy_gff_filename=self.good_fuzzy_gff_fn,
fuzzy_group_filename=self.fuzzy_group_fn,
allow_extra_5exon=self.allow_extra_5exon,
max_fuzzy_junction=self.max_fuzzy_junction)
logging.info("Good fuzzy isoforms written to: %s", realpath(self.good_fuzzy_gff_fn))
logging.info("Bad fuzzy isoforms written to: %s", realpath(self.bad_fuzzy_gff_fn))
logging.info("Fuzzy isoform groups written to: %s", realpath(self.fuzzy_group_fn))
ln(self.good_fuzzy_gff_fn, self.good_gff_fn)
ln(self.good_fuzzy_gff_fn, self.gff_fn)
ln(self.fuzzy_group_fn, self.group_fn)
else:
logging.info("No need to further collapse fuzzy junctions.")
ln(self.good_unfuzzy_gff_fn, self.good_gff_fn)
ln(self.good_unfuzzy_gff_fn, self.gff_fn)
ln(self.unfuzzy_group_fn, self.group_fn)
# Pick up representative
logging.info("Picking up representative record.")
pick_least_err_instead = not self.allow_extra_5exon # 5merge, pick longest
pick_rep(isoform_filename=self.isoform_filename,
gff_filename=self.good_gff_fn,
group_filename=self.group_fn,
output_filename=self.rep_fn(self.suffix),
pick_least_err_instead=pick_least_err_instead,
bad_gff_filename=self.bad_gff_fn)
logging.info("Ignored IDs written to: %s", realpath(self.ignored_ids_txt_fn))
logging.info("Output GFF written to: %s", realpath(self.gff_fn))
logging.info("Output Group TXT written to: %s", realpath(self.group_fn))
logging.info("Output collapsed isoforms written to: %s", realpath(self.rep_fn(self.suffix)))
logging.info("CollapseIsoforms Arguments: %s", self.arg_str())
def run(self):
"""
First, collapse input isoforms by calling Branch.run().
Then collapse fuzzy junctions by calling collapse_fuzzy_junctions.
Finally, pick up representitive gff record for each group of collapsed isoforms.
"""
self.validate_inputs()
logging.info("Collapsing isoforms into transcripts.")
b = Branch(isoform_filename=self.isoform_filename,
sam_filename=self.sam_filename,
cov_threshold=self.min_flnc_coverage,
min_aln_coverage=self.min_aln_coverage,
min_aln_identity=self.min_aln_identity)
b.run(allow_extra_5exon=self.allow_extra_5exon,
skip_5_exon_alt=self.skip_5_exon_alt,
ignored_ids_fn=self.ignored_ids_txt_fn,
good_gff_fn=self.good_unfuzzy_gff_fn,
bad_gff_fn=self.bad_unfuzzy_gff_fn,
group_fn=self.unfuzzy_group_fn)
logging.info("Good unfuzzy isoforms written to: %s",
realpath(self.good_unfuzzy_gff_fn))
logging.info("Bad unfuzzy isoforms written to: %s",
realpath(self.bad_unfuzzy_gff_fn))
logging.info("Unfuzzy isoform groups written to: %s",
realpath(self.unfuzzy_group_fn))
if self.shall_collapse_fuzzy_junctions:
logging.info("Further collapsing fuzzy junctions.")
# need to further collapse those that have fuzzy junctions!
collapse_fuzzy_junctions(
gff_filename=self.good_unfuzzy_gff_fn,
group_filename=self.unfuzzy_group_fn,
fuzzy_gff_filename=self.good_fuzzy_gff_fn,
fuzzy_group_filename=self.fuzzy_group_fn,
allow_extra_5exon=self.allow_extra_5exon,
max_fuzzy_junction=self.max_fuzzy_junction)
logging.info("Good fuzzy isoforms written to: %s",
realpath(self.good_fuzzy_gff_fn))
logging.info("Bad fuzzy isoforms written to: %s",
realpath(self.bad_fuzzy_gff_fn))
logging.info("Fuzzy isoform groups written to: %s",
realpath(self.fuzzy_group_fn))
ln(self.good_fuzzy_gff_fn, self.good_gff_fn)
ln(self.good_fuzzy_gff_fn, self.gff_fn)
ln(self.fuzzy_group_fn, self.group_fn)
else:
logging.info("No need to further collapse fuzzy junctions.")
ln(self.good_unfuzzy_gff_fn, self.good_gff_fn)
ln(self.good_unfuzzy_gff_fn, self.gff_fn)
ln(self.unfuzzy_group_fn, self.group_fn)
# Pick up representative
logging.info("Picking up representative record.")
pick_least_err_instead = not self.allow_extra_5exon # 5merge, pick longest
pick_rep(isoform_filename=self.isoform_filename,
gff_filename=self.good_gff_fn,
group_filename=self.group_fn,
output_filename=self.rep_fn(self.suffix),
pick_least_err_instead=pick_least_err_instead,
bad_gff_filename=self.bad_gff_fn)
logging.info("Ignored IDs written to: %s",
realpath(self.ignored_ids_txt_fn))
logging.info("Output GFF written to: %s", realpath(self.gff_fn))
logging.info("Output Group TXT written to: %s",
realpath(self.group_fn))
logging.info("Output collapsed isoforms written to: %s",
realpath(self.rep_fn(self.suffix)))
logging.info("CollapseIsoforms Arguments: %s", self.arg_str())