def CheckSuboptimal(rep_id, exons, eliminated_predictions, other_ids,
map_prediction2data, options):
overlaps = []
# get predictions which overlap by exons (but not completely):
for id in other_ids:
if id == rep_id:
continue
if id in eliminated_predictions:
continue
if Exons.CheckOverlap( exons[rep_id], exons[id]) and \
not Exons.CheckCoverage(exons[rep_id],
exons[id],
max_slippage=options.max_slippage):
overlaps.append(id)
rep = map_prediction2data[rep_id]
identity = rep.mPid + options.suboptimal_min_identity_difference
for x in range(0, len(overlaps) - 1):
id1 = overlaps[x]
d1 = map_prediction2data[id1]
for y in range(x + 1, len(overlaps)):
id2 = overlaps[y]
d2 = map_prediction2data[id2]
if options.loglevel >= 3:
options.stdlog.write(
"# suboptimal: %s ? %s + %s: %s %s %s %s %i %i %i\n" % (
rep_id,
id1,
id2,
d1.mQuality in options.quality_remove_suboptimal,
d2.mQuality in options.quality_remove_suboptimal,
not Exons.CheckOverlap(exons[id1], exons[id2]),
Exons.CheckCoverageAinB(
exons[rep_id],
exons[id1] + exons[id2],
min_terminal_exon_coverage=0.0),
rep.mPid,
d1.mPid,
d2.mPid,
))
if (d1.mQuality in options.quality_remove_suboptimal and
d2.mQuality in options.quality_remove_suboptimal ) and \
not Exons.CheckOverlap( exons[id1], exons[id2] ) and \
Exons.CheckContainedAinB(exons[rep_id], exons[id1] + exons[id2],
min_terminal_exon_coverage=0.0 ) and \
(identity < d1.mPid) and \
(identity < d2.mPid):
if options.loglevel >= 1:
options.stdlog.write(
"# elimination: %s(%s) joins %s(%s) and %s(%s)\n" %
(rep_id, rep.mPid, id1, d1.mPid, id2, d2.mPid))
return True
return False
def CheckExonSwop( rep_id,
exons,
eliminated_predictions,
other_ids,
map_prediction2data,
options ):
"""check for exon swop
return true, if exon swop occurs.
Exon swop occurs, if this prediction joins
two predictions, one of which should be CG.
None of the predictions should be fully contained
in the master prediction.
given:
the rep_id to analyzse
a map of rep_id to exons
a list of rep_ids to check against
-> is it an exon swopper?
-> joining two CG predictions that do not overlap and
contain no extra exons apart from the overlapping.
-> is it large spanning prediction?
-> spanning many predictions, including at least one CG?
"""
overlaps = []
## get predictions which overlap by exons (but not completely):
for id in other_ids:
if id == rep_id: continue
if id in eliminated_predictions: continue
if Exons.CheckOverlap( exons[rep_id], exons[id]) and \
not Exons.CheckCoverage( exons[rep_id],
exons[id],
max_slippage=options.max_slippage ):
overlaps.append( id )
if options.loglevel >= 3:
options.stdlog.write( "# exon swop: %s overlaps with %i out of %i predictions\n" % (rep_id, len(overlaps), len(other_ids) ) )
options.stdlog.flush()
for x in range(0, len(overlaps)-1):
id1 = overlaps[x]
for y in range(x+1, len(overlaps)):
id2 = overlaps[y]
if options.loglevel >= 4:
options.stdlog.write( "# exon swop: %s ? %s + %s: %s %s %s %s\n" % \
(rep_id, id1, id2,
map_prediction2data[id1].mQuality in options.quality_remove_exon_swopper,
map_prediction2data[id2].mQuality in options.quality_remove_exon_swopper,
not Exons.CheckOverlap( exons[id1], exons[id2] ),
Exons.CheckCoverageAinB( exons[rep_id], exons[id1] + exons[id2],
min_terminal_num_exons = 0,
min_terminal_exon_coverage = 0.7,
max_slippage = options.max_slippage ) ) )
if (map_prediction2data[id1].mQuality in options.quality_remove_exon_swopper and \
map_prediction2data[id2].mQuality in options.quality_remove_exon_swopper ) and \
not Exons.CheckOverlap( exons[id1], exons[id2] ) and \
Exons.CheckCoverageAinB( exons[rep_id], exons[id1] + exons[id2],
min_terminal_num_exons = 0,
min_terminal_exon_coverage = 0.7,
max_slippage = options.max_slippage ):
if options.loglevel >= 1:
options.stdlog.write( "# elimination: %s(%s) joins %s(%s) and %s(%s)\n" % \
(rep_id, map_prediction2data[rep_id].mQuality,
id1, map_prediction2data[id1].mQuality,
id2, map_prediction2data[id2].mQuality) )
return True
return False