def process(matches):
new = matches[0].copy()
map_query2target = alignlib_lite.py_makeAlignmentBlocks()
graph = networkx.DiGraph()
graph.add_nodes_from(range(len(matches) + 2))
matches.sort(key=lambda x: x.mQueryFrom)
if Genomics.IsPositiveStrand(matches[0].strand):
f = lambda x, y: x.mSbjctTo < y.mSbjctFrom
else:
f = lambda x, y: x.mSbjctFrom > y.mSbjctTo
for x in range(0, len(matches)):
xx = matches[x]
if options.loglevel >= 6:
options.stdlog.write("# graph: %2i %s\n" % (x, str(xx)))
for y in range(x + 1, len(matches)):
yy = matches[y]
d = min(xx.mQueryTo, yy.mQueryTo) - \
max(xx.mQueryFrom, yy.mQueryFrom)
if d > 0 or not f(xx, yy):
continue
else:
graph.add_edge(x, y, {'weight': -d})
source = len(matches)
target = len(matches) + 1
for x in range(len(matches)):
xx = matches[x]
graph.add_edge(source, x, {'weight': xx.mQueryFrom})
graph.add_edge(
x, target, {'weight': xx.mQueryLength - xx.mQueryTo})
if options.loglevel >= 6:
networkx.write_edgelist(graph, options.stdlog)
path = networkx.dijkstra_path(graph, source, target)
if options.loglevel >= 6:
options.stdlog.write("# path: %s\n" % (str(path)))
new_matches = [matches[x] for x in path[1:-1]]
if len(matches) != len(new_matches):
E.warn(("query=%s, target=%s, strand=%s: "
"removed overlapping/out-of-order segments: "
"before=%i, after=%i") %
(matches[0].mQueryId,
matches[0].mSbjctId,
matches[0].strand,
len(matches),
len(new_matches)))
matches = new_matches
for match in matches:
m = match.getMapQuery2Target()
alignlib_lite.py_addAlignment2Alignment(map_query2target, m)
new.fromMap(map_query2target, use_strand=True)
options.stdout.write(str(new) + "\n")
options.stdout.flush()
return 1
def Add(self,
const_other,
combine_contig=False,
allow_overlap=False,
contig_size=0,
combine_queries=False,
as_intron=False):
"""add one entry to another.
This procedure allows to add
- predictions on different contigs if combine_contig = True
- overlapping predictions on the same query if allow_overlap = True
- results from different queries if combine_queries = True
- if as_intron is set to true, the new fragment is added as an intron.
"""
# create working copies of each prediction
other = const_other.getCopy()
this = self.getCopy()
other.Expand()
this.Expand()
if as_intron:
code = "I"
else:
code = "P"
# check for query overlaps
if this.mQueryToken == other.mQueryToken:
query_overlap = max(
0,
min(this.mQueryTo, other.mQueryTo) -
max(this.mQueryFrom, other.mQueryFrom) + 1)
if query_overlap > 0:
if allow_overlap:
overlap = query_overlap
# if queries overlap, truncate this before adding the other
this.mMapPeptide2Translation.removeRowRegion(
this.mQueryTo - overlap + 1, this.mQueryTo)
other.mMapPeptide2Translation.moveAlignment(0, -overlap)
this.mQueryTo -= overlap
this.mTranslation = this.mTranslation[:-overlap]
# remove aligned residues from the back
for x in range(len(this.mMapPeptide2Genome) - 1, 0, -1):
if this.mMapPeptide2Genome[x][1] <= overlap:
overlap -= this.mMapPeptide2Genome[x][1]
del this.mMapPeptide2Genome[x]
else:
break
this.mMapPeptide2Genome[-1] = (
this.mMapPeptide2Genome[-1][0],
this.mMapPeptide2Genome[-1][1] - overlap,
this.mMapPeptide2Genome[-1][2] - overlap * 3)
else:
raise ValueError, "refusing to add overlapping entries: overlap = %i, queries:\n%s\n%s\n, set allow_overlap = True " % (
query_overlap, str(this), str(other))
else:
if not combine_queries:
raise ValueError, "refusing to add different queries - set combine_queries = True."
if this.mSbjctToken != other.mSbjctToken or \
this.mSbjctStrand != other.mSbjctStrand:
if combine_contig:
this.mSbjctToken += "-" + other.mSbjctToken
this.mSbjctStrand += other.mSbjctStrand
else:
raise ValueError, "can not add different sbjct."
sbjct_overlap = max(
0,
min(this.mSbjctGenomeTo, other.mSbjctGenomeTo) -
max(this.mSbjctGenomeFrom, other.mSbjctGenomeFrom), 0)
if sbjct_overlap > 0:
if not combine_contig:
raise ValueError, "refusing to add overlapping entries: overlap = %i, sbjct:\n%s\n%s\n" % (
sbjct_overlap, str(this), str(other))
if this.mSbjctToken == other.mSbjctToken:
# set precedence
if this.mSbjctGenomeFrom < other.mSbjctGenomeFrom:
first = this
second = other
else:
first = other
second = this
# get length of gap
d_na = second.mSbjctGenomeFrom - first.mSbjctGenomeTo
if this.mQueryToken != other.mQueryToken:
d_aa = first.mQueryLength - first.mQueryTo
# create a new virtual query by concatenating
# the two queries
this.mQueryToken += "-" + other.mQueryToken
# sort out the alignment
second.mMapPeptide2Translation.moveAlignment(
first.mQueryLength, 0)
this.mQueryLength = first.mQueryLength + second.mQueryLength
else:
d_aa = second.mQueryFrom - first.mQueryTo - 1
this.mSbjctGenomeFrom = min(this.mSbjctGenomeFrom,
other.mSbjctGenomeFrom)
this.mSbjctGenomeTo = max(this.mSbjctGenomeTo,
other.mSbjctGenomeTo)
this.mMapPeptide2Genome = first.mMapPeptide2Genome + \
[(code, d_aa, d_na)] + second.mMapPeptide2Genome
this.mTranslation = first.mTranslation + second.mTranslation
second.mMapPeptide2Translation.moveAlignment(0, first.mSbjctTo - 1)
else:
# join on different contigs
d_na = contig_size - this.mSbjctGenomeTo + \
other.mSbjctGenomeFrom + query_overlap * 3
d_aa = other.mQueryFrom - this.mQueryTo - 1
this.mMapPeptide2Genome += [(code, d_aa, d_na), ] + \
other.mMapPeptide2Genome
this.mTranslation += other.mTranslation
other.mMapPeptide2Translation.moveAlignment(0, this.mSbjctTo - 1)
this.mSbjctGenomeFrom = this.mSbjctGenomeFrom
this.mSbjctGenomeTo = contig_size + other.mSbjctGenomeTo
# now fill self from first and this
self.mQueryToken = first.mQueryToken
self.mQueryLength = this.mQueryLength
nthis = this.mMapPeptide2Translation.getLength(
) - this.mMapPeptide2Translation.getNumGaps()
nother = other.mMapPeptide2Translation.getLength(
) - other.mMapPeptide2Translation.getNumGaps()
self.mMapPeptide2Genome = first.mMapPeptide2Genome
self.mSbjctGenomeFrom = this.mSbjctGenomeFrom
self.mSbjctGenomeTo = this.mSbjctGenomeTo
# there might be some reference counting issues, thus
# do it the explicit way.
alignlib_lite.py_addAlignment2Alignment(this.mMapPeptide2Translation,
other.mMapPeptide2Translation)
self.mMapPeptide2Translation = alignlib_lite.py_makeAlignmentVector()
alignlib_lite.py_addAlignment2Alignment(self.mMapPeptide2Translation,
this.mMapPeptide2Translation)
self.mTranslation = this.mTranslation
self.mQueryFrom = self.mMapPeptide2Translation.getRowFrom()
self.mQueryTo = self.mMapPeptide2Translation.getRowTo()
self.mSbjctFrom = self.mMapPeptide2Translation.getColFrom()
self.mSbjctTo = self.mMapPeptide2Translation.getColTo()
self.mQueryCoverage = 100.0 * \
(self.mQueryTo - self.mQueryFrom + 1) / float(self.mQueryLength)
self.mAlignmentString = string.join(
map(lambda x: string.join(map(str, x), " "),
self.mMapPeptide2Genome), " ")
f = alignlib_lite.py_AlignmentFormatEmssions(
self.mMapPeptide2Translation)
self.mQueryAli, self.mSbjctAli = f.mRowAlignment, f.mColAlignment
# summary parameters
self.mRank = max(this.mRank, other.mRank)
self.score += other.score
self.mNGaps += other.mNGaps
self.mNFrameShifts += other.mNFrameShifts
self.mNIntrons += other.mNIntrons + 1
self.mNStopCodons += other.mNStopCodons
nnew = self.mMapPeptide2Translation.getLength(
) - self.mMapPeptide2Translation.getNumGaps()
self.mPercentIdentity = min(
100.0,
(self.mPercentIdentity * nthis + other.mPercentIdentity * nother) /
nnew)
self.mPercentSimilarity = min(
100.0, (self.mPercentSimilarity * nthis +
other.mPercentSimilarity * nother) / nnew)
self.mNAssembled += 1 + other.mNAssembled
def process(matches):
new = matches[0].copy()
map_query2target = alignlib_lite.py_makeAlignmentBlocks()
graph = networkx.DiGraph()
graph.add_nodes_from(xrange(len(matches) + 2))
matches.sort(key=lambda x: x.mQueryFrom)
if Genomics.IsPositiveStrand(matches[0].strand):
f = lambda x, y: x.mSbjctTo < y.mSbjctFrom
else:
f = lambda x, y: x.mSbjctFrom > y.mSbjctTo
for x in range(0, len(matches)):
xx = matches[x]
if options.loglevel >= 6:
options.stdlog.write("# graph: %2i %s\n" % (x, str(xx)))
for y in range(x + 1, len(matches)):
yy = matches[y]
d = min(xx.mQueryTo, yy.mQueryTo) - \
max(xx.mQueryFrom, yy.mQueryFrom)
if d > 0 or not f(xx, yy):
continue
else:
graph.add_edge(x, y, {'weight': -d})
source = len(matches)
target = len(matches) + 1
for x in range(len(matches)):
xx = matches[x]
graph.add_edge(source, x, {'weight': xx.mQueryFrom})
graph.add_edge(
x, target, {'weight': xx.mQueryLength - xx.mQueryTo})
if options.loglevel >= 6:
networkx.write_edgelist(graph, options.stdlog)
path = networkx.dijkstra_path(graph, source, target)
if options.loglevel >= 6:
options.stdlog.write("# path: %s\n" % (str(path)))
new_matches = [matches[x] for x in path[1:-1]]
if len(matches) != len(new_matches):
E.warn(("query=%s, target=%s, strand=%s: "
"removed overlapping/out-of-order segments: "
"before=%i, after=%i") %
(matches[0].mQueryId,
matches[0].mSbjctId,
matches[0].strand,
len(matches),
len(new_matches)))
matches = new_matches
for match in matches:
m = match.getMapQuery2Target()
alignlib_lite.py_addAlignment2Alignment(map_query2target, m)
new.fromMap(map_query2target, use_strand=True)
options.stdout.write(str(new) + "\n")
options.stdout.flush()
return 1
def Add( self, const_other,
combine_contig = False,
allow_overlap = False,
contig_size = 0,
combine_queries = False,
as_intron = False ):
"""add one entry to another.
This procedure allows to add
- predictions on different contigs if combine_contig = True
- overlapping predictions on the same query if allow_overlap = True
- results from different queries if combine_queries = True
- if as_intron is set to true, the new fragment is added as an intron.
"""
## create working copies of each prediction
other = const_other.getCopy()
this = self.getCopy()
other.Expand()
this.Expand()
if as_intron:
code = "I"
else:
code = "P"
## check for query overlaps
if this.mQueryToken == other.mQueryToken:
query_overlap = max( 0, min(this.mQueryTo, other.mQueryTo) -\
max(this.mQueryFrom, other.mQueryFrom) + 1)
if query_overlap > 0:
if allow_overlap:
overlap = query_overlap
## if queries overlap, truncate this before adding the other
this.mMapPeptide2Translation.removeRowRegion( this.mQueryTo - overlap + 1, this.mQueryTo )
other.mMapPeptide2Translation.moveAlignment( 0, -overlap )
this.mQueryTo -= overlap
this.mTranslation = this.mTranslation[:-overlap]
## remove aligned residues from the back
for x in range(len(this.mMapPeptide2Genome) - 1, 0, -1):
if this.mMapPeptide2Genome[x][1] <= overlap:
overlap -= this.mMapPeptide2Genome[x][1]
del this.mMapPeptide2Genome[x]
else:
break
this.mMapPeptide2Genome[-1] = (this.mMapPeptide2Genome[-1][0],
this.mMapPeptide2Genome[-1][1] - overlap,
this.mMapPeptide2Genome[-1][2] - overlap * 3)
else:
raise ValueError, "refusing to add overlapping entries: overlap = %i, queries:\n%s\n%s\n, set allow_overlap = True " % (query_overlap, str(this), str(other))
else:
if not combine_queries:
raise ValueError, "refusing to add different queries - set combine_queries = True."
if this.mSbjctToken != other.mSbjctToken or \
this.mSbjctStrand != other.mSbjctStrand :
if combine_contig:
this.mSbjctToken += "-" + other.mSbjctToken
this.mSbjctStrand += other.mSbjctStrand
else:
raise ValueError, "can not add different sbjct."
sbjct_overlap = max(0, min(this.mSbjctGenomeTo, other.mSbjctGenomeTo) -\
max(this.mSbjctGenomeFrom, other.mSbjctGenomeFrom), 0)
if sbjct_overlap > 0:
if not combine_contig:
raise ValueError, "refusing to add overlapping entries: overlap = %i, sbjct:\n%s\n%s\n" % (sbjct_overlap, str(this), str(other))
if this.mSbjctToken == other.mSbjctToken:
## set precedence
if this.mSbjctGenomeFrom < other.mSbjctGenomeFrom:
first = this
second = other
else:
first = other
second = this
## get length of gap
d_na = second.mSbjctGenomeFrom - first.mSbjctGenomeTo
if this.mQueryToken != other.mQueryToken:
d_aa = first.mQueryLength - first.mQueryTo
# create a new virtual query by concatenating
# the two queries
this.mQueryToken += "-" + other.mQueryToken
# sort out the alignment
second.mMapPeptide2Translation.moveAlignment( first.mQueryLength, 0 )
this.mQueryLength = first.mQueryLength + second.mQueryLength
else:
d_aa = second.mQueryFrom - first.mQueryTo - 1
this.mSbjctGenomeFrom = min(this.mSbjctGenomeFrom, other.mSbjctGenomeFrom )
this.mSbjctGenomeTo = max(this.mSbjctGenomeTo, other.mSbjctGenomeTo )
this.mMapPeptide2Genome = first.mMapPeptide2Genome + [(code, d_aa, d_na)] + second.mMapPeptide2Genome
this.mTranslation = first.mTranslation + second.mTranslation
second.mMapPeptide2Translation.moveAlignment( 0, first.mSbjctTo - 1 )
else:
## join on different contigs
d_na = contig_size - this.mSbjctGenomeTo + other.mSbjctGenomeFrom + query_overlap * 3
d_aa = other.mQueryFrom - this.mQueryTo - 1
this.mMapPeptide2Genome += [(code, d_aa, d_na),] + other.mMapPeptide2Genome
this.mTranslation += other.mTranslation
other.mMapPeptide2Translation.moveAlignment( 0, this.mSbjctTo - 1 )
this.mSbjctGenomeFrom = this.mSbjctGenomeFrom
this.mSbjctGenomeTo = contig_size + other.mSbjctGenomeTo
## now fill self from first and this
self.mQueryToken = first.mQueryToken
self.mQueryLength = this.mQueryLength
nthis = this.mMapPeptide2Translation.getLength() - this.mMapPeptide2Translation.getNumGaps()
nother = other.mMapPeptide2Translation.getLength() - other.mMapPeptide2Translation.getNumGaps()
self.mMapPeptide2Genome = first.mMapPeptide2Genome
self.mSbjctGenomeFrom = this.mSbjctGenomeFrom
self.mSbjctGenomeTo= this.mSbjctGenomeTo
## there might be some reference counting issues, thus
## do it the explicit way.
alignlib_lite.py_addAlignment2Alignment( this.mMapPeptide2Translation, other.mMapPeptide2Translation)
self.mMapPeptide2Translation = alignlib_lite.py_makeAlignmentVector()
alignlib_lite.py_addAlignment2Alignment( self.mMapPeptide2Translation, this.mMapPeptide2Translation )
self.mTranslation = this.mTranslation
self.mQueryFrom = self.mMapPeptide2Translation.getRowFrom()
self.mQueryTo = self.mMapPeptide2Translation.getRowTo()
self.mSbjctFrom = self.mMapPeptide2Translation.getColFrom()
self.mSbjctTo = self.mMapPeptide2Translation.getColTo()
self.mQueryCoverage = 100.0 * (self.mQueryTo - self.mQueryFrom + 1) / float(self.mQueryLength)
self.mAlignmentString = string.join( map( \
lambda x: string.join(map(str, x), " "),
self.mMapPeptide2Genome), " ")
f = alignlib_lite.py_AlignmentFormatEmssions( self.mMapPeptide2Translation )
self.mQueryAli, self.mSbjctAli = f.mRowAlignment, f.mColAlignment
## summary parameters
self.mRank = max( this.mRank, other.mRank)
self.score += other.score
self.mNGaps += other.mNGaps
self.mNFrameShifts += other.mNFrameShifts
self.mNIntrons += other.mNIntrons + 1
self.mNStopCodons += other.mNStopCodons
nnew = self.mMapPeptide2Translation.getLength() - self.mMapPeptide2Translation.getNumGaps()
self.mPercentIdentity = min( 100.0, (self.mPercentIdentity * nthis + other.mPercentIdentity * nother) / nnew )
self.mPercentSimilarity = min( 100.0, (self.mPercentSimilarity * nthis + other.mPercentSimilarity * nother) / nnew )
self.mNAssembled += 1 + other.mNAssembled