def load_gff(gff):
genes = defaultdict(list)
gene_exon_positions = defaultdict(lambda: defaultdict(tuple))
try:
with open(gff) as g:
for line in g:
if line.startswith('#') or 'contig' in line:
continue
feature = GFF(line)
gene_id = get_gene_attribute(feature, "ID")
if feature.featuretype == 'exon':
gene_exon_positions[feature.genename][gene_id] = (
feature.start, feature.end)
if feature.featuretype == 'CDS':
for exon in gene_exon_positions[feature.genename]:
e = gene_exon_positions[feature.genename][exon]
if e[0] <= feature.start <= e[1] and e[
0] <= feature.end <= e[1]:
gene_id = exon + "_CDS"
if gene_id is None:
print("No gene id for CDS found", feature, end="")
feature.id = gene_id
genes[feature.genename].append(feature)
except IOError:
print("Failed to load GFF file {}".format(gff))
sys.exit()
return genes
def load_crossmap(crossmapout):
genes = defaultdict(list)
gene_exon_positions = defaultdict(lambda: defaultdict(tuple))
try:
with open(crossmapout) as c:
for line in c:
cm = None
if 'fail' in line:
featureline = line.split('\tfail')[0]
if '->' in line:
featureline, cmline = line.split('\t->\t')
cm = GFF(cmline)
cm.attributes += ';Note=CrossMap'
feature = GFF(featureline)
feature.crossmap = cm
gene_id = get_gene_attribute(feature, "ID")
if feature.featuretype == 'exon':
gene_exon_positions[feature.genename][gene_id] = (
feature.start, feature.end)
if feature.featuretype == 'CDS':
for exon in gene_exon_positions[feature.genename]:
e = gene_exon_positions[feature.genename][exon]
if e[0] <= feature.start <= e[1] and e[
0] <= feature.end <= e[1]:
gene_id = exon + "_CDS"
if feature.crossmap: # Lose CrossMap features where the CDSs aren't the same length
cds_len = feature.end - feature.start + 1
cm_len = feature.crossmap.end - feature.crossmap.start + 1
if cds_len != cm_len:
feature.crossmap = None
if gene_id is None:
print("No gene id for CDS found", feature, end="")
feature.id = gene_id
genes[feature.genename].append(feature)
except IOError:
print("Failed to load CrossMap output {}".format(crossmapout))
sys.exit()
return genes
def make_output_feature(outstart, outend, feature, new_parts):
if outstart is not None and outend is not None:
if outstart > outend:
outstart, outend = outend, outstart
strand = feature.strand
if new_parts[0].strand == -1:
if feature.strand == '+':
strand = '-'
elif feature.strand == '-':
strand = '+'
return GFF(new_parts[0].oldname, feature.source, feature.featuretype,
outstart, outend, feature.score, strand, feature.phase,
feature.attributes)
else:
return None
def transfer_gff_feature(feature, genome, crossmap=False):
output_feature = outstart = outend = status = None
if crossmap and feature.crossmap:
output_feature = GFF(feature.crossmap.scaffold,
feature.crossmap.source,
feature.crossmap.featuretype,
feature.crossmap.start, feature.crossmap.end,
feature.crossmap.score, feature.crossmap.strand,
feature.crossmap.phase,
feature.crossmap.attributes)
status = 'crossmap'
return output_feature, status
new_parts, haps = get_parts(feature.scaffold, feature.start, feature.end,
genome)
if not new_parts:
if not haps:
status = 'missing'
else:
status = haps[0].parttype
elif len(new_parts) == 1 and not haps:
np = new_parts[0]
if 'removed' not in np.parttype:
outstart, outend = new_parts[0].oldstart, new_parts[0].oldend
status = 'ok'
else:
status = 'removed'
else:
scaffolds = {}
for part in new_parts:
if 'removed' in part.parttype:
continue
if part.newstart <= feature.start <= part.newend:
outstart = part.oldstart if part.strand == 1 else part.oldend
scaffolds[part.oldname] = 1
if part.newstart <= feature.end <= part.newend:
outend = part.oldend if part.strand == 1 else part.oldstart
scaffolds[part.oldname] = 1
if outstart is not None and outend is not None:
status = 'ok'
if len(scaffolds) > 1:
outstart = outend = None
status = 'multiscaffold'
if feature.featuretype == 'CDS': # Do not allow CDSs to span multiple parts, but try to find a CrossMap hit
outstart = outend = None
status = 'broken'
output_feature = make_output_feature(outstart, outend, feature, new_parts)
if output_feature is None and feature.crossmap:
output_feature = GFF(feature.crossmap.scaffold,
feature.crossmap.source,
feature.crossmap.featuretype,
feature.crossmap.start, feature.crossmap.end,
feature.crossmap.score, feature.crossmap.strand,
feature.crossmap.phase,
feature.crossmap.attributes)
status = 'crossmap'
if output_feature is None:
if status is None:
status = 'broken'
return GFF(feature.scaffold, feature.source, feature.featuretype,
feature.start, feature.end, feature.score, feature.strand,
feature.phase, feature.attributes), status
else:
return output_feature, status
