def profileMapLengths(self):
cmapLens = []
nMaps = 0
totalLen = 0
for curCmap in self.cmapDB.values():
cmapLen = curCmap.cmapLen
cmapLens.append(cmapLen)
totalLen += cmapLen
nMaps += 1
if nMaps:
self.n = nMaps
self.n50 = util.getn50(cmapLens)
self.totalLen = totalLen
self.aveLen = totalLen / nMaps
def profileMapLengths(self):
cmapLens = []
nMaps = 0
totalLen = 0
for curCmap in self.cmapDB.values():
cmapLen = curCmap.cmapLen
cmapLens.append(cmapLen)
totalLen += cmapLen
nMaps += 1
if nMaps:
self.n = nMaps
self.n50 = util.getn50(cmapLens)
self.totalLen = totalLen
self.aveLen = totalLen / nMaps
def characterizeContigs(varsP, xmappath=None):
"""Log simple contigs stats, and optionally align stats from xmappath.
"""
#print "xmappath:", xmappath
unitscale = 1e-6
dorefalign = bool(
xmappath
) #i'm never actually calling refaligner here--this is just using xmappath
haveref = bool(varsP.ref)
#refcmap = mapClasses.multiCmap() #not used
aligndir = varsP.contigAlignTarget
try:
#refcmap = mapClasses.multiCmap(varsP.ref)
#reflen = refcmap.totalLength #note: total length of _all_ contigs
reflen = mapClasses.multiCmap(varsP.ref, lengthonly=True).totalLength
#in summary table, this is a denominator--make sure it's non-zero, don't bail (still get contig summary)
if reflen <= 0:
#print "Warning in CharacterizeModule.characterizeContigs: bad reflen", reflen, "defaulting to 1" #not necessary
reflen = 1.
except:
reflen = 1.
outstr = "" #Contig Characterization:\n"
#check for .hmaps in same dir as latestMergedCmap: if any, add a line for haploid genome size
hmaps = util.getListOfFilesFromDir(os.path.dirname(varsP.latestMergedCmap),
".hmap")
haplotype = (len(hmaps) > 0)
haplotypelen = 0
hapcontiglens = []
totcontiglen = 0
totalignlen = 0
nmapcontigs = 0
totalignqlen = 0 #defalignlen = 0;
contiglens = [] #lens of all contigs in bases
uniqueseg = {
} #the argument to util.uniqueRange--stores all the map lengths which go into totalignlen--now each of these is a value, and the keys are the reference contig id or chromosome depending on dorefidchr
for citr, cpath in enumerate([varsP.latestMergedCmap
]): #always use contigpaths
mapi = mapClasses.multiCmap(cpath)
totcontiglen += mapi.totalLength
contiglens += mapi.getAllMapLengths(
) #getAllMapLengths is list of all map lengths
if haplotype:
haplotypelen += mapi.getHaplotypeTotalMapLength()
hapcontiglens.extend(mapi.getHaplotypeMapLengths())
#store a list of the contig ids in this multiCmap, then remove them if they're in the xmap
# if they're not, print at the end
mapids = mapi.getAllMapIds(
) #this is once per cmap, not once per characterizeModule call--becuase it's keys, it's already a copy, no need to copy explicitly
ncontigs = len(
mapids) #this is ncontigs in this file, ie, in mapi (see below)
xmapobj = mapClasses.xmap() #empty map to fix xmapobj scope
if dorefalign: #get xmap object
if util.checkFile(xmappath, ".xmap"):
xmapobj = mapClasses.xmap(xmappath)
for xitr, xmapentry in enumerate(xmapobj.xmapLookup.values()):
#get map length from multicmap.getMapLength--returns 0 for any exception
contiglen = mapi.getMapLength(xmapentry.contigQry)
if contiglen <= 0: #this strikes me as clumsy...but I don't want to return non-zero from multiCmap.getMapLength
contiglen = 1.
contigcov = mapi.getMapAvgCoverage(xmapentry.contigQry)
#don't print lenr for each contig--just total them
lenr = xmapentry.getMappedRefLen()
lenq = xmapentry.getMappedQryLen()
refid = xmapentry.contigRef #int
totalignlen += lenr
totalignqlen += lenq
#uniqueseg is now a dict to take into account which chromosome the query contig is on
#note need refid bc need to separate different contigs on the _same_ chromosome
if not uniqueseg.has_key(
refid
): #if first contig on chromosome, need to init new list
uniqueseg[refid] = []
uniqueseg[refid].append([xmapentry.RefStart, xmapentry.RefStop])
#process mapids--remove contig id (contigQry) from mapids if they're in the xmap so non-aligning contigs can be printed
if xmapentry.contigQry in mapids:
mapids.remove(xmapentry.contigQry)
#end loop on xmap entries
#now that all xmap entries are processed, all contigs with an alignment are removed from mapids,
# so we can get n contigs align using this and ncontigs
nmapcontigs += ncontigs - len(mapids) #sum multiple cmaps
#end loop on contigs
varsP.totAssemblyLenMb = totcontiglen * unitscale
ncontigs = len(
contiglens) #contigpaths is just files--contiglens is all contigs
avgcontiglen = (float(totcontiglen) / ncontigs if ncontigs > 0 else 0)
if unitscale > 1e-6: #if not megabases
fstr = "%9.0f"
else: #megabases
fstr = "%8.3f"
if haplotype: #new format for haplotype
#if haplotypelen != sum(hapcontiglens) : #simply print warning in this case (do not log): ignore this bc of floating point rounding
#print "Warning in characterizeContigs: haplotype lengths are inconsistent:", haplotypelen, sum(hapcontiglens)
#diploid is same as else below, but names change
outstr += "Diploid N Genome Maps: %i\n" % ncontigs
outstr += ("Diploid Genome Map Len (Mb): " + fstr +
"\n") % (totcontiglen * unitscale)
outstr += ("Diploid Avg. Genome Map Len (Mb): " + fstr +
"\n") % (avgcontiglen * unitscale)
outstr += ("Diploid Median Genome Map Len (Mb): " + fstr +
"\n") % (util.getMedian(contiglens) * unitscale)
outstr += ("Diploid Genome Map n50 (Mb): " + fstr +
"\n") % (util.getn50(contiglens) * unitscale)
#haploid : ignore haplotypelen, just use the list hapcontiglens
outstr += "Haploid N Genome Maps: %i\n" % len(hapcontiglens)
tot = sum(hapcontiglens)
avg = (tot / len(hapcontiglens) if len(hapcontiglens) else 0)
outstr += ("Haploid Genome Map Len (Mb): " + fstr +
"\n") % (tot * unitscale)
outstr += ("Haploid Avg. Genome Map Len (Mb): " + fstr +
"\n") % (avg * unitscale)
outstr += ("Haploid Median Genome Map Len (Mb): " + fstr +
"\n") % (util.getMedian(hapcontiglens) * unitscale)
outstr += ("Haploid Genome Map n50 (Mb): " + fstr +
"\n") % (util.getn50(hapcontiglens) * unitscale)
else: #default to old format
outstr += "N Genome Maps: %i\n" % ncontigs
outstr += ("Total Genome Map Len (Mb): " + fstr +
"\n") % (totcontiglen * unitscale)
outstr += ("Avg. Genome Map Len (Mb): " + fstr +
"\n") % (avgcontiglen * unitscale)
outstr += ("Median Genome Map Len (Mb): " + fstr +
"\n") % (util.getMedian(contiglens) * unitscale)
outstr += ("Genome Map n50 (Mb): " + fstr +
"\n") % (util.getn50(contiglens) * unitscale)
if haveref:
outstr += ("Total Ref Len (Mb): " + fstr + "\n") % (reflen *
unitscale)
outstr += ("Total Genome Map Len / Ref Len : " + fstr +
"\n") % (totcontiglen / reflen)
if dorefalign:
ratio = (float(nmapcontigs) / ncontigs if ncontigs > 0 else 0)
outstr += ("N Genome Maps total align : %i (%.2f)\n") % (
nmapcontigs, ratio)
outstr += ("Total Aligned Len (Mb) : " + fstr +
"\n") % (totalignlen * unitscale)
outstr += ("Total Aligned Len / Ref Len : " + fstr +
"\n") % (totalignlen / reflen)
uniquelen = 0
for segs in uniqueseg.values(): # need to sum on dict entries
util.uniqueRange(segs) #this modifies list in place
uniquelen += util.totalLengthFromRanges(segs)
outstr += ("Total Unique Aligned Len (Mb) : " + fstr +
"\n") % (uniquelen * unitscale)
outstr += ("Total Unique Aligned Len / Ref Len: " + fstr +
"\n") % (uniquelen / reflen)
return outstr
def characterizeContigs(varsP, xmappath=None, listcontigs=False) :
"""Log simple contigs stats, and optionally align stats from xmappath.
"""
#print header of table
unitscale = 1e-6
dorefalign = bool(xmappath) #i'm never actually calling refaligner here--this is just using xmappath
#dorefidchr = False
#dorefcid = False
printrange = False
#printsegrms = False
#dochrstr = False
iscluster = True
haveref = bool(varsP.ref)
#refcmap = mapClasses.multiCmap() #not used
aligndir = varsP.contigAlignTarget
try :
#refcmap = mapClasses.multiCmap(varsP.ref)
#reflen = refcmap.totalLength #note: total length of _all_ contigs
reflen = mapClasses.multiCmap(varsP.ref, lengthonly=True).totalLength
#in summary table, this is a denominator--make sure it's non-zero, don't bail (still get contig summary)
if reflen <= 0 :
#print "Warning in CharacterizeModule.characterizeContigs: bad reflen", reflen, "defaulting to 1" #not necessary
reflen = 1.
except:
reflen = 1.
outstr = "" #Contig Characterization:\n"
if listcontigs and dorefalign :
outstr += "cID len"
outstr += " Cov"
#if dorefidchr or dorefcid :
# outstr += " rID" #ref index for either of these
#if dorefidchr :
# outstr += " rpos"
outstr += " alignlen alignlen/len"
if printrange :
outstr += " Qry1 Qry2 Ref1 Ref2"
#outstr += (" segRMS" if printsegrms else "")
outstr += " Conf Conf/lenkb"
outstr += " FP FN sf sd bpp" #" res" #--ditch res (not bpp)
#if dochrstr :
# outstr += " Chr"
outstr += "\n"
totcontiglen = 0; totalignlen = 0; nmapcontigs = 0; defalignlen = 0; totalignqlen = 0
contiglens = [] #lens of all contigs in bases
uniqueseg = {} #the argument to util.uniqueRange--stores all the map lengths which go into totalignlen--now each of these is a value, and the keys are the reference contig id or chromosome depending on dorefidchr
avgfplist = []; avgfnlist = [] #average FP/FN rates
#if dorefidchr :
# chrsum = refcmap.makeChrSummaryDict() #see mapClasses.multiCmap
for citr, cpath in enumerate([varsP.latestMergedCmap]) : #always use contigpaths
mapi = mapClasses.multiCmap(cpath)
totcontiglen += mapi.totalLength
contiglens += mapi.getAllMapLengths() #getAllMapLengths is list of all map lengths
#store a list of the contig ids in this multiCmap, then remove them if they're in the xmap
# if they're not, print at the end
mapids = mapi.getAllMapIds() #this is once per cmap, not once per characterizeModule call--becuase it's keys, it's already a copy, no need to copy explicitly
ncontigs = len(mapids) #this is ncontigs in this file, ie, in mapi (see below)
xmapobj = mapClasses.xmap() #empty map to fix xmapobj scope
if dorefalign : #get xmap object
#xmappath = aligndir+os.path.split(cpath)[-1].replace(".cmap", ".xmap") #need cmap file name
#xmappath = self.xmapTarget
#if xmappath exists isn't a file, nothing will be loaded
if os.path.isfile( xmappath ) : #was if not isfile : continue
xmapobj = mapClasses.xmap(xmappath)
for xitr, xmapentry in enumerate(xmapobj.xmapLookup.values()) :
#print the contig id from the xmap
#this is sorted by ref position; could sort the list this loop is over by the contigQry data member,
# _but_, I think I like reference-oriented better because you can see gap spanning
#get map length from multicmap.getMapLength--returns 0 for any exception
contiglen = mapi.getMapLength(xmapentry.contigQry)
if contiglen <= 0 : #this strikes me as clumsy...but I don't want to return non-zero from multiCmap.getMapLength
contiglen = 1.
contigcov = mapi.getMapAvgCoverage(xmapentry.contigQry)
if listcontigs :
outstr += "%5i" % xmapentry.contigQry
outstr += " %9.1f %2i" % (contiglen, contigcov)
#don't print lenr for each contig--just total them
lenr = xmapentry.getMappedRefLen()
lenq = xmapentry.getMappedQryLen()
refid = xmapentry.contigRef #int
#if dorefidchr : #this is the encoding of ref contig id to chromosome and start position
# chrpos = mapClasses.intToChrPos(refid, verbose=False) #this returns a tuple (shouldn't fail, but verbose false)
# refidstr = " %2s %6i" % chrpos
# chrs = chrpos[0] #just for readability
# if chrsum.has_key(chrs) : #the method that fills chrsum (makeChrSummaryDict) also uses intToChrPos
# chrsum[chrs][0] += lenr #values are list, first ele is total aligned length, second is ref length (fixed)
#elif dorefcid :
# refidstr = " %3i" % refid #refid is int
#else : #nothing for neither, but still need empty string
refidstr = ""
conf = xmapentry.Confidence #confidence from the xmap, and ratio of it to length in kb
if listcontigs :
alignpars = getMappedErrStats(aligndir, cpath) #an empty err file is produced for case of no align
avgfplist.append( alignpars.fp )
avgfnlist.append( alignpars.fn )
outstr += "%s %9.1f %.3f" % (refidstr, lenq, lenq/contiglen) #length for refidstr set above
if printrange :
outstr += " %5.0f %5.0f %5.0f %5.0f" % (xmapentry.QryStart/pn, xmapentry.QryStop/pn, xmapentry.RefStart/pn, xmapentry.RefStop/pn)
#outstr += (" %5.0f" % 0 if printsegrms else "") #don't print anything
outstr += " %3.0f %5.3f" % (conf, conf*1000./lenq) #1000 is for kb
outstr += " " + alignpars.getParamString()
totalignlen += lenr
totalignqlen += lenq
#uniqueseg is now a dict to take into account which chromosome the query contig is on
#note need refid bc need to separate different contigs on the _same_ chromosome
if not uniqueseg.has_key(refid) : #if first contig on chromosome, need to init new list
uniqueseg[refid] = []
uniqueseg[refid].append( [xmapentry.RefStart, xmapentry.RefStop] )
#process mapids--remove contig id (contigQry) from mapids if they're in the xmap so non-aligning contigs can be printed
if xmapentry.contigQry in mapids :
mapids.remove(xmapentry.contigQry)
#note: the feature of multiple alignments (strict vs default) is no longer implemented
defalignlen += lenr #currently, just default and strict
#if listcontigs and dochrstr :
# outstr += " " + refIndexToChrStr( xmapentry.contigRef )
# outstr += "\n"
#end loop on xmap entries
#now that all xmap entries are processed, all contigs with an alignment are removed from mapids,
# so we can get n contigs align using this and ncontigs
nmapcontigs += ncontigs - len(mapids) #sum multiple cmaps
#and print the data for the contigs which don't align--just id, length, and coverage
#these lines are kind of redundant, but I guess that's ok
if listcontigs :
for ids in mapids :
outstr += "%5i" % ids
#get map length from multicmap.getMapLength--returns 0 for any exception
contiglen = mapi.getMapLength(ids) #it's ok if it's 0 bc it's never a denominator here
contigcov = mapi.getMapAvgCoverage(ids)
outstr += " %9.1f %2i\n" % (contiglen, contigcov)
#end loop on contigs
varsP.totAssemblyLenMb = totcontiglen*unitscale
ncontigs = len(contiglens) #contigpaths is just files--contiglens is all contigs
avgcontiglen = (float(totcontiglen)/ncontigs if ncontigs > 0 else 0)
#print averages
if listcontigs and not iscluster : #only do avg if not merged, otherwise just one noise parameter
avgfp = sum(avgfplist)/len(avgfplist)
avgfn = sum(avgfnlist)/len(avgfnlist)
outstr += "AVG %9.1f %9.1f %5.3f %5.3f\n" % (avgcontiglen, totalignqlen/nmapcontigs, avgfp, avgfn)
if unitscale > 1e-6 : #if not megabases
fstr = "%9.0f"
else : #megabases
fstr = "%8.3f"
outstr += "N Genome Maps: %i\n" % ncontigs
outstr += ("Total Genome Map Len (Mb): "+fstr+"\n") % (totcontiglen*unitscale)
outstr += ("Avg. Genome Map Len (Mb): "+fstr+"\n") % (avgcontiglen*unitscale)
outstr += ("Median Genome Map Len(Mb): "+fstr+"\n") % (util.getMedian(contiglens)*unitscale)
outstr += ("Genome Map n50 (Mb): "+fstr+"\n") % (util.getn50(contiglens)*unitscale)
if haveref :
outstr += ("Total Ref Len (Mb): "+fstr+"\n") % (reflen*unitscale)
outstr += ("Total Genome Map Len / Ref Len : "+fstr+"\n") % (totcontiglen/reflen)
if dorefalign :
#print the chromosome summary before the strict/default/total align stats
#if dorefidchr :
# outstr += "Chromosome Summary:\n"
# outstr += "Chromosome align len ref len (ratio):\n"
# for chrs, align in chrsum.iteritems() :
# outstr += "%3s %9.0f %9.0f (%5.3f)\n" % (chrs, align[0], align[1], align[0]/align[1])
ratio = (float(nmapcontigs)/ncontigs if ncontigs > 0 else 0)
outstr += ("N Genome Maps total align : %i (%.2f)\n") % (nmapcontigs, ratio)
outstr += ("Total Aligned Len (Mb) : "+fstr+"\n") % (totalignlen*unitscale)
outstr += ("Total Aligned Len / Ref Len : "+fstr+"\n") % (totalignlen/reflen)
uniquelen = 0
for segs in uniqueseg.values() : # need to sum on dict entries
util.uniqueRange(segs) #this modifies list in place
uniquelen += util.totalLengthFromRanges( segs )
outstr += ("Total Unique Aligned Len (Mb) : "+fstr+"\n") % (uniquelen*unitscale)
outstr += ("Total Unique Aligned Len / Ref Len: "+fstr+"\n") % (uniquelen/reflen)
return outstr
