def getGeneTable(reffile):
E.info("Loading reference")
table = defaultdict(dict)
for ens_gene in GTF.gene_iterator(GTF.iterator(
IOTools.open_file(reffile))):
geneid = ens_gene[0][0].gene_id
table[geneid]["models"] = dict()
table[geneid]["start_codons"] = defaultdict(list)
for transcript in ens_gene:
transcript_id = transcript[0].transcript_id
table[geneid]["models"][transcript_id] = transcript
CDS = GTF.asRanges(transcript, "start_codon")
if len(CDS) == 0:
continue
if transcript[0].strand == "-":
start_codon = max(e[1] for e in CDS)
else:
start_codon = min(e[0] for e in CDS)
table[geneid]["start_codons"][start_codon].append(transcript_id)
E.info("Reference Loaded")
return table
def transcript2bed12(transcript):
new_entry = Bed.Bed()
start = min(entry.start for entry in transcript)
end = max(entry.end for entry in transcript)
try:
thickStart = min(entry.start for entry in transcript
if entry.feature == "CDS")
thickEnd = max(entry.end for entry in transcript
if entry.feature == "CDS")
except ValueError:
# if there is no CDS, then set first base of transcript as
# start
if transcript[0].strand == "-":
thickStart = end
thickEnd = end
else:
thickStart = start
thickEnd = start
exons = GTF.asRanges(transcript, "exon")
exon_starts = [es - start for (es, ee) in exons]
exon_lengths = [ee - es for (es, ee) in exons]
exon_count = len(exons)
new_entry.contig = transcript[0].contig
new_entry.start = start
new_entry.end = end
new_entry["strand"] = transcript[0].strand
new_entry["name"] = transcript[0].transcript_id
new_entry["thickStart"] = thickStart
new_entry["thickEnd"] = thickEnd
new_entry["blockCount"] = exon_count
new_entry["blockStarts"] = ",".join(map(str, exon_starts))
new_entry["blockSizes"] = ",".join(map(str, exon_lengths))
return new_entry
def main(argv=None):
"""script main.
parses command line options in sys.argv, unless *argv* is given.
"""
if argv is None:
argv = sys.argv
# setup command line parser
parser = E.OptionParser(version="%prog version: $1.0$",
usage=globals()["__doc__"])
parser.add_option("-r",
"--reffile",
dest="reffile",
type="string",
help="Supply reference gtf file name")
parser.add_option("-d",
"--class-file",
dest="classfile",
type="string",
help="Supply database name")
parser.add_option("-o",
"--outfile",
dest="outfile",
type="string",
help="Supply output bed file name")
parser.add_option("-u",
"--indivfile",
dest="indivfile",
type="string",
help="Supply output bed file name for individual utrons")
parser.add_option("-p",
"--partfile",
dest="partfile",
type="string",
help="Supply output bed file name for partnered utrons")
parser.add_option(
"-q",
"--indivpartfile",
dest="indivpartfile",
type="string",
help="Supply output bed file name for individual partnered utrons")
parser.add_option("-n",
"--novel-file",
dest="novelfile",
type="string",
help="Supply output bed file name for novel introns")
parser.add_option(
"--novel-transcript",
dest="novel_id",
type="string",
help="DEBUG: Output info for this transcript from the STDIN")
parser.add_option(
"--target-transcript",
dest="target_id",
type="string",
help="DEBUG: Output info for this transcript from ref-file")
# add common options (-h/--help, ...) and parse command line
(options, args) = E.start(parser, argv=argv)
outlines = []
individuals = []
partnered = []
individualpartnered = []
novel = []
db = pandas.read_csv(options.classfile, sep="\t")
# This keeps just one entry per-transcript - why?
#db = db.groupby("transcript_id").first()
db = db.set_index("transcript_id")
enshashtable = getGeneTable(options.reffile)
for novel_transcript in GTF.transcript_iterator(GTF.iterator(
options.stdin)):
# Why do it on a gene by gene basis rather than transcript by transcript basis?
transcript_id = novel_transcript[0].transcript_id
if transcript_id == options.novel_id:
output_novel = True
else:
output_novel = False
try:
geneid = db.loc[transcript_id].match_gene_id
except KeyError:
if output_novel:
E.debug("Transcript %s not in class table" % transcript_id)
continue
if pandas.isnull(geneid):
if output_novel:
E.debug("Transcript %s matches no gene in class table" %
transcript_id)
continue
ens_gene = enshashtable[geneid]
all_ref_introns = set()
novel_transcript_exons = GTF.asRanges(novel_transcript, "exon")
novel_transcript_introns = GTF.toIntronIntervals(novel_transcript)
for ref_transcript in ens_gene["models"].values():
ref_introns = GTF.toIntronIntervals(ref_transcript)
all_ref_introns.update(ref_introns)
#Identify comparison set
def _in_exon(position, exons):
return any(e[0] <= position <= e[1] for e in exons)
# check if this ever gets the wrong start_codon.
filtered_starts = [
s for s in ens_gene["start_codons"]
if _in_exon(s, novel_transcript_exons)
]
if len(filtered_starts) == 0:
if output_novel:
E.debug("No starts found for %s" % transcript_id)
continue
#if novel_transcript[0].strand == "-":
# selected_start = max(filtered_starts)
#else:
# selected_start = min(filtered_starts)
selected_models = list()
for startc in filtered_starts:
selected_models.extend(ens_gene["start_codons"][startc])
if output_novel:
E.debug("Transcripts with compatible starts are %s" %
selected_models)
for ref_transcript_id in selected_models:
if output_novel and ref_transcript_id == options.target_id:
output_ref = True
else:
output_ref = False
second = ens_gene["models"][ref_transcript_id]
ens_CDS = GTF.asRanges(second, "CDS")
if len(ens_CDS) == 0:
if output_ref:
E.debug("%s is not coding"
) # ensure only protein-coding transcripts
continue
ens_exons = GTF.asRanges(second, "exon")
first_introns = set(novel_transcript_introns)
second_introns = set(GTF.toIntronIntervals(second))
first_CDSintrons = [
intron for intron in first_introns
if (intron[0] > ens_CDS[0][0] and intron[1] < ens_CDS[-1][1])
]
second_CDSintrons = [
intron for intron in second_introns
if (intron[0] > ens_CDS[0][0] and intron[1] < ens_CDS[-1][1])
]
first_CDSintrons = set(first_CDSintrons)
second_CDSintrons = set(second_CDSintrons)
if not first_CDSintrons == second_CDSintrons:
if output_ref:
E.debug("CDS chains do not match. Chains are:")
first_CDSintrons = sorted(list(first_CDSintrons))
second_CDSintrons = sorted(list(second_CDSintrons))
output = "\n".join(
map(str, zip(first_CDSintrons, second_CDSintrons)))
E.debug(output)
continue # match CDS intron chain
firstUTRintrons = first_introns - first_CDSintrons
if len(firstUTRintrons) == 0:
if output_ref:
E.debug("No UTR introns")
continue
secondUTRintrons = second_introns - second_CDSintrons
found = False
for intron in first_introns:
if (intron[0] < ens_CDS[-1][1] and
intron[1] > ens_CDS[-1][1]) or \
(intron[0] < ens_CDS[0][0] and
intron[1] > ens_CDS[0][0]):
found = True
break # ensure pruned transcript doesn't have
# introns overlapping start or stop codons in ensembl
# transcript
if found:
if output_ref:
E.debug("Start or stop in intron")
continue
if second[0].strand == "+":
ens_stop = ens_CDS[-1][1]
UTR3introns = [
intron for intron in firstUTRintrons
if intron[0] >= ens_CDS[-1][1]
and intron[1] < ens_exons[-1][1]
]
secondUTR3introns = [
intron for intron in secondUTRintrons
if intron[0] >= ens_CDS[-1][1]
and intron[1] < ens_exons[-1][1]
]
else:
ens_stop = ens_CDS[0][0]
UTR3introns = [
intron for intron in firstUTRintrons if
intron[1] <= ens_CDS[0][0] and intron[0] > ens_exons[0][0]
]
secondUTR3introns = [
intron for intron in secondUTRintrons if
intron[1] <= ens_CDS[0][0] and intron[0] > ens_exons[0][0]
]
if len(UTR3introns) == 0:
if output_ref:
E.debug("No UTR introns")
continue
outbed = Bed.Bed()
outbed.fields = ['.', '.', '.', '.', '.', '.', '.', '.', '.']
outbed.fromIntervals(UTR3introns)
outbed.contig = novel_transcript[0].contig
outbed["name"] = novel_transcript[0].transcript_id
outbed["strand"] = novel_transcript[0].strand
outlines.append(outbed) # get output for each transcript
for item in UTR3introns:
outbed2 = Bed.Bed()
outbed2.fields = ['.', '.', '.', '.']
outbed2.fromIntervals([item])
outbed2.contig = novel_transcript[0].contig
outbed2['name'] = novel_transcript[0].transcript_id
outbed2["strand"] = novel_transcript[0].strand
outbed2["thickStart"] = ens_stop
individuals.append(outbed2) # get output for each intron
UTR3introns = set(UTR3introns)
secondUTR3introns = set(secondUTR3introns)
extraUTR3introns = list(UTR3introns - secondUTR3introns)
if output_ref and len(secondUTR3introns - UTR3introns) > 0:
E.debug("Following introns in UTR of %s but not %s" %
(options.target_id, options.novel_id))
E.debug(secondUTRintrons - UTR3introns)
# get only introns that are not in matched transcript
if len(extraUTR3introns) != 0 and len(secondUTR3introns -
UTR3introns) == 0:
outbed3 = Bed.Bed()
outbed3.fields = ['.'] * 9
outbed3.fromIntervals(extraUTR3introns)
outbed3.contig = novel_transcript[0].contig
outbed3["name"] = novel_transcript[
0].transcript_id + ":" + second[0].transcript_id
outbed3["strand"] = novel_transcript[0].strand
partnered.append(outbed3)
for item in extraUTR3introns:
outbed4 = Bed.Bed()
outbed4.fields = ['.', '.', '.', '.']
outbed4.fromIntervals([item])
outbed4.contig = novel_transcript[0].contig
outbed4["name"] = novel_transcript[
0].transcript_id + ":" + second[0].transcript_id
outbed4["strand"] = novel_transcript[0].strand
outbed4["thickStart"] = ens_stop
individualpartnered.append(outbed4)
if len(all_ref_introns) == 0:
ens_starts, ens_ends = [], []
else:
ens_starts, ens_ends = zip(*all_ref_introns)
novelEvents = [
i for i in UTR3introns
if i[0] not in ens_starts and i[1] not in ens_ends
]
for item in novelEvents:
outbed5 = Bed.Bed()
outbed5.fields = ['.'] * 4
outbed5.fromIntervals([item])
outbed5.contig = novel_transcript[0].contig
outbed5["name"] = novel_transcript[
0].transcript_id + ":" + second[0].transcript_id
outbed5["strand"] = novel_transcript[0].strand
outbed5["thickStart"] = ens_stop
novel.append(outbed5)
with IOTools.open_file(options.outfile, "w") as outf:
for line in outlines:
outf.write(str(line) + "\n")
if options.indivfile is not None:
with IOTools.open_file(options.indivfile, "w") as outf2:
for line in individuals:
outf2.write(str(line) + "\n")
if options.partfile is not None:
with IOTools.open_file(options.partfile, "w") as outf3:
for line in partnered:
outf3.write(str(line) + "\n")
if options.indivpartfile is not None:
with IOTools.open_file(options.indivpartfile, "w") as outf4:
for line in individualpartnered:
outf4.write(str(line) + "\n")
if options.novelfile is not None:
with IOTools.open_file(options.novelfile, "w") as outf5:
for line in novel:
outf5.write(str(line) + "\n")
# write footer and output benchmark information.
E.stop()
