def flagExonStatus(infile, outfile):
'''
Adds two attributes to the gtf entry:
exon_status_locus - specifies whether the gene model is multi- or
single-exon
exon_status - specifies whether the transcript is mult- or single exon
'''
PipelineLncRNA.flagExonStatus(infile, outfile)
def buildRefcodingGeneSetStats(infile, outfile):
'''
counts:
no. of transcripts
no. genes
average number of exons per transcript
average number of exons per gene
no. multi-exon transcripts
no. single exon transcripts
no. multi-exon genes
no. single exon genes
in the coding and lncRNA genesets
'''
# calculate exon status for refcoding genes.
tmpf = P.getTempFilename(".") + ".gz"
PipelineLncRNA.flagExonStatus(infile, tmpf)
outf = iotools.openFile(outfile, "w")
outf.write("\t".join([
"no_transcripts", "no_genes", "no_exons_per_transcript",
"no_exons_per_gene", "no_single_exon_transcripts",
"no_multi_exon_transcripts", "no_single_exon_genes",
"no_multi_exon_genes"
]) + "\n")
outf.write("\t".join(
map(str, [
PipelineLncRNA.CounterTranscripts(tmpf).count(),
PipelineLncRNA.CounterGenes(tmpf).count(),
PipelineLncRNA.CounterExonsPerTranscript(tmpf).count(),
PipelineLncRNA.CounterExonsPerGene(tmpf).count(),
PipelineLncRNA.CounterSingleExonTranscripts(tmpf).count(),
PipelineLncRNA.CounterMultiExonTranscripts(tmpf).count(),
PipelineLncRNA.CounterSingleExonGenes(tmpf).count(),
PipelineLncRNA.CounterMultiExonGenes(tmpf).count()
])))
os.unlink(tmpf)
os.unlink(tmpf + ".log")
os.unlink(P.snip(tmpf, ".gz"))