def classifyLncRNA(infiles, outfile):
'''
Classify lncRNA realtive to protein coding loci
Classify lincRNA in terms of their relationship to
protein coding genes - creates indices for intervals on the
fly - mayb should be creating additional annotations:
antisense
transcript overlapping protein coding exons on opposite strand
antisense_upstream
transcript < 2kb from tss on opposite strand
antisense_downstream
transcript < 2kb from gene end on opposite strand
sense_upstream
transcript < 2kb from tss on same strand
sense_downstream
transcript < 2kb from gene end on same strand
intergenic
transcript >2kb from any protein coding gene
intronic
overlaps protein coding gene intron on same strand
antisense_intronic
overlaps protein coding intron on opposite strand
'''
PipelineLncRNA.classifyLncRNAGenes(infiles[0],
infiles[1],
outfile,
dist=PARAMS["lncrna_dist"])
def classifyFilteredLncRNA(infiles, outfile):
'''
classifies all lincRNA before cpc filtering to define any classes that
are represented in the coding set that are filtered
NOTE: This task is not included when running the full pipeline
'''
PipelineLncRNA.classifyLncRNAGenes(infiles[0],
infiles[1],
outfile,
dist=PARAMS["lncrna_dist"])