def __init__(
self,
gff_filename: Union[str, Path],
group_filename: Union[str, Path],
internal_fuzzy_max_dist: int = 0,
self_prefix: str = None,
fastq_filename: Union[str, Path] = None,
fusion_max_dist: int = 10,
):
"""
Differences with non-fusion MegaPBTree:
1. allow_5merge is always FALSE. Not a parameter.
2. fusion_max_dist --- maximum allowed distance on internal fusion sites to be called as equivalent fusions
"""
super().__init__(
gff_filename,
group_filename,
internal_fuzzy_max_dist,
self_prefix,
False,
fastq_filename,
)
self.fusion_max_dist = fusion_max_dist
# ex: PBfusion.1 -> [PBfusion.1.1, PBfusion.1.2]
self.record_d_fusion = {
fusion_id: records
for fusion_id, records in GFF.collapseGFFFusionReader(gff_filename)
}
def add_sample(
self,
gff_filename: Union[str, Path],
group_filename: Union[str, Path],
sample_prefix: str,
output_prefix: str,
fastq_filename: Optional[Union[str, Path]] = None,
) -> None:
combined = (
[]
) # list of (r1 if r2 is None | r2 if r1 is None | longer of r1 or r2 if both not None)
unmatched_recs = list(self.record_d_fusion.keys())
for _, records in GFF.collapseGFFFusionReader(gff_filename):
match_seqid = self.match_fusion_record(records)
if match_seqid is not None:
combined.append((self.record_d_fusion[match_seqid], records))
try:
unmatched_recs.remove(match_seqid)
except ValueError:
pass # already deleted, OK, this happens for single-exon transcripts
else: # r is not present in current tree
combined.append((None, records))
# put whatever is left from the tree in
for seqid in unmatched_recs:
combined.append((self.record_d_fusion[seqid], None))
# create a ClusterTree to re-calc the loci/transcripts
final_tree = defaultdict(lambda: {
"+": ClusterTree(0, 0),
"-": ClusterTree(0, 0)
})
for i, (r1s, r2s) in enumerate(combined):
if r2s is None or (r1s is not None and r1s[0].end - r1s[0].start >
r2s[0].end - r2s[0].start):
final_tree[r1s[0].chr][r1s[0].strand].insert(
r1s[0].start, r1s[0].end, i)
else:
final_tree[r2s[0].chr][r2s[0].strand].insert(
r2s[0].start, r2s[0].end, i)
self.write_cluster_tree_as_gff(
final_tree,
combined,
group_filename,
sample_prefix,
output_prefix,
fastq_filename2=fastq_filename,
)
def sample_sanity_check(
group_filename: Union[str, Path],
gff_filename: Union[str, Path],
count_filename: Union[str, Path],
fastq_filename: Optional[Union[str, Path]] = None,
) -> None:
"""
Double check that the formats are expected and all PBIDs are concordant across the files
:return: raise Exception if sanity check failed
"""
logger.info(
f"Sanity checking. Retrieving PBIDs from {group_filename},{gff_filename},{count_filename}..."
)
ids1 = [line.strip().split()[0] for line in open(group_filename)]
ids2 = [
fusion_id
for fusion_id, rs in GFF.collapseGFFFusionReader(gff_filename)
]
with open(count_filename) as f:
for _ in range(14):
f.readline() # just through the header
ids3 = [r["pbid"] for r in DictReader(f, delimiter="\t")]
if len({ids2}.difference(ids1)) > 0 or len({ids2
}.difference(ids3)) > 0:
raise Exception(
f"Sanity check failed! Please make sure the PBIDs listed in {gff_filename} are also in {gff_filename} and {count_filename}"
)
if fastq_filename is not None:
ids4 = [
r.id.split("|")[0] for r in SeqIO.parse(fastq_filename, "fastq")
]
if len({ids2}.difference(ids4)) > 0:
raise Exception(
f"Sanity check failed! Please make sure the PBIDs listed in {gff_filename} are also in {fastq_filename}"
)