def LoadExonBoundaryCoordinatesFromFile(self, path): #{
DebugMsg(self, "Gene annotation file: %s" % path)
# open the annotations file
annotations_file = GeneAnnotationParserCls(path, log_info=self.log_info)
skipped_chroms = set()
# get the coordinates from the file
for transcript in annotations_file: #{
# fix chromosome names, if needed
chrom = NormalizeChrID(transcript.chrom)
if (NonStandardChr(chrom)): #{
ExtremeDebugMsg(self, "Skipping transcript in strange chromosome: "
"%s (%s)" % (chrom, transcript.chrom))
skipped_chroms.add(chrom)
continue
#} end if
prev_exon = None
for (index, exon) in enumerate(transcript.SortedExons()): #{
(exon.left, exon.right) = (exon.min, exon.max)
# assume that exon list is sorted by left coordinate
if (None != prev_exon and prev_exon.min > exon.min): #{
raise ExonBoundCounterError("Transcript %s exons are not in "
"order: %s, %s" % (transcript.transcript_id,
prev_exon.ToString(), exon.ToString()))
#} end if
prev_exon = exon
# do not include the left side of the first exon
if (0 == index): #{
exon.left = None
#} end if
# do not include the right side of the last exon
if (len(transcript.exons) == (index+1)): #{
exon.right = None
#} end if
# exon_bound_coords[chrom][prime_side][coord1][coord2] = gene_list
for side in SIDES: #{
if (None != getattr(exon, side)): #{
keys = [chrom, side, getattr(exon, side),
getattr(exon, OtherSide(side))]
AddToMultiDict(self.exon_bound_coords, keys,
transcript.transcript_id)
#} end if
#} end for
#} end for
#} end for
if (0 < len(skipped_chroms)): #{
DebugMsg(self, "Skipped transcripts in chromosomes: %s" %
", ".join(sorted(skipped_chroms)))
#} end if
# close the file
annotations_file.close()
