def averagemotifs(motifs,ovlp=2,template=None,DFUNC=negcommonbitsrange,VERBOSE=1,prop=''):
if not template:
Dmat = computeDmat(motifs)
idx = centroididx(Dmat)
template = motifs[idx]
for m in motifs:
off, rc = minshortestoverhangdiff(template,m,OVLP(template,m),'want_offset',DFUNC=DFUNC)
m.offset = off
m.rc = rc
#Find most negative offset
offsets = [m.offset for m in motifs] ; offsets.sort()
maxposs = [(m.offset + m.width) for m in motifs] ; maxposs.sort()
minpos = -offsets[0]
maxpos = maxposs[-1] + minpos
pmotifs = []
for m in motifs:
if m.rc: _m = m.revcomp()
else : _m = m
leftpad = minpos + m.offset
rightpad = maxpos - (leftpad + m.width)
padded = _m[-leftpad,_m.width+rightpad]
#print '%s%s%s\t%s'%('*'*leftpad,_m.oneletter,'*'*rightpad,padded)
pmotifs.append(padded)
AVE = MotifTools.sum(pmotifs,[])
if VERBOSE:
for m in pmotifs:
d = minshortestoverhangdiff(AVE,m,OVLP(AVE,m),DFUNC=DFUNC)
print '%s %5.3f'%(m.oneletter,d),
if m.__dict__.has_key('key'): print m.key,
if prop and m.__dict__.has_key(prop): print m.__dict__[prop],
print
print '-'*m.width
return AVE
def alignAndCombineMotifs(motifs, weights):
# zip motifs and weights
simMotifs = zip(motifs, weights)
# sort by weights
simMotifs.sort(key=lambda x: abs(x[1]))
simMotifs.reverse()
aligned = alignSimilarMotifs([x[0] for x in simMotifs], minoverlap=4)
#print '--'
#for each in aligned: print each.oneletter
#print '\n'
comboMotif = MotifTools.sum(aligned,[-x[1] for x in simMotifs])
return comboMotif
def averagemotifs(motifs,
ovlp=2,
template=None,
DFUNC=negcommonbitsrange,
VERBOSE=1,
prop=''):
if not template:
Dmat = computeDmat(motifs)
idx = centroididx(Dmat)
template = motifs[idx]
for m in motifs:
off, rc = minshortestoverhangdiff(template,
m,
OVLP(template, m),
'want_offset',
DFUNC=DFUNC)
m.offset = off
m.rc = rc
#Find most negative offset
offsets = [m.offset for m in motifs]
offsets.sort()
maxposs = [(m.offset + m.width) for m in motifs]
maxposs.sort()
minpos = -offsets[0]
maxpos = maxposs[-1] + minpos
pmotifs = []
for m in motifs:
if m.rc: _m = m.revcomp()
else: _m = m
leftpad = minpos + m.offset
rightpad = maxpos - (leftpad + m.width)
padded = _m[-leftpad, _m.width + rightpad]
#print '%s%s%s\t%s'%('*'*leftpad,_m.oneletter,'*'*rightpad,padded)
pmotifs.append(padded)
AVE = MotifTools.sum(pmotifs, [])
if VERBOSE:
for m in pmotifs:
d = minshortestoverhangdiff(AVE, m, OVLP(AVE, m), DFUNC=DFUNC)
print '%s %5.3f' % (m.oneletter, d),
if m.__dict__.has_key('key'): print m.key,
if prop and m.__dict__.has_key(prop): print m.__dict__[prop],
print
print '-' * m.width
return AVE
# TAMOify kmers and logify pVals
for i in range(len(testMotifs)):
testMotifs[i] = (Motif(testMotifs[i][0]),numpy.log10(float(testMotifs[i][1])))
# Sort on log'd pVals
testMotifs.sort(key=lambda x: x[1])
comboMotifs = []
for i in range(0,int(len(testMotifs)*0.2)):
simMotifs = getKmersWithOneMisMtch(testMotifs[i][0],testMotifs)
alndMotifs = alignSimilarMotifs([x[0] for x in simMotifs])
#for m in simMotifs:
#print m[0].oneletter
comboMotifs.append(MotifTools.sum(alndMotifs,[-x[1] for x in simMotifs])) # -x[1] to convert neg logs to pos weights
print len(comboMotifs)
t2 = time.time()
oFile = '/Users/biggus/Documents/James/Collaborations/Campbell/data/Results_HyperGeoScreen/masked/Results_gGEMS/CCupAt4Days.gte2x.5-16mers.shfSeq.3.gGEMS.tmo'
pFile = '/Users/biggus/Documents/James/Collaborations/Campbell/data/Results_HyperGeoScreen/masked/Results_gGEMS/CCupAt4Days.gte2x.5-16mers.shfSeq.3.gGEMS.pkl'
MotifTools.save_motifs(comboMotifs,oFile,kmer_count=60)
pFile = open(pFile, 'w')
cPickle.dump(comboMotifs,pFile)
t3 = time.time()
print 'Calculations took %.3f min.\nWriting/Pickling took %.3f min.' % ((float(t2)-t1)/60, (float(t3)-t2)/60)
from TAMO import MotifTools
motifs = ['TGATACA',
'TGATAAA',
'TGACAAA',
'AGATACA',
'AGATACG',]
for i in range(len(motifs)):
motifs[i] = MotifTools.Motif_from_text(motifs[i])
weights = [3,
7,
2,
2,
1,]
noWeight = MotifTools.sum(motifs)
withWeights = MotifTools.sum(motifs, weights)
print "nW:\n%s\n" % (noWeight.printlogo())
print "wW:\n%s" % (withWeights.printlogo())
x=1
if mtf.width > lMotifLen:
lMotifLen = mtf.width
for i in range(len(alignedMotifs)):
if alignedMotifs[i].width < lMotifLen:
alignedMotifs[i] = alignedMotifs[i][0,lMotifLen]
return alignedMotifs
m = MotifTools.load('/Users/biggus/Documents/James/Collaborations/Campbell/data/Results_HyperGeoScreen/masked/Results_gGEMS/CCupAt4Days.6-8mers.gGEMS.top6.motifs.stdThresh.tmo')
w = [5.8952,
5.6523,
5.0585,
4.9788,
4.9678,
4.7688]
twoFive = [[m[0],m[1],m[4]],[w[0],w[1],w[4]]]
alndMotifs = alignSimilarMotifs(twoFive[0], minoverlap=4)
for m in alndMotifs:
print m.oneletter
sumdMotif = MotifTools.sum(alndMotifs)
#bKmers = sumdMotif.bogus_kmers()
#for k in bKmers:
#print k
None
