def main():
seqsD = Fasta.load(sys.argv[1])
seqs = seqsD.values()
for w in range(1, 7):
allnmers = permute(w)
nmersT = MotifTools.top_nmers(w, seqs, 'with counts', 'purge Ns')
nmersD = {}
total = 0
for nmer in allnmers:
nmersD[nmer] = 1 #Pseudo count
total = total + 1
for nmer, count in nmersT[:]:
try:
rc = MotifTools.revcomplement(nmer)
nmersD[nmer] = nmersD[nmer] + count
nmersD[rc] = nmersD[rc] + count
total = total + 2 * count
except KeyError:
pass
_t = nmersD.keys()
_t.sort()
print "# freq in %s (total %d with pseudocounts)" % (sys.argv[1],
total)
for nmer in _t:
print "%-7s %20.17f" % (nmer, float(nmersD[nmer]) / total)
sys.stdout.flush()
def main():
seqsD = Fasta.load(sys.argv[1])
seqs = seqsD.values()
for w in range(1,7):
allnmers = permute(w)
nmersT = MotifTools.top_nmers(w,seqs,'with counts','purge Ns')
nmersD = {}
total = 0
for nmer in allnmers:
nmersD[nmer] = 1 #Pseudo count
total = total + 1
for nmer,count in nmersT[:]:
try:
rc = MotifTools.revcomplement(nmer)
nmersD[nmer] = nmersD[nmer] + count
nmersD[rc] = nmersD[rc] + count
total = total + 2*count
except KeyError:
pass
_t = nmersD.keys()
_t.sort()
print "# freq in %s (total %d with pseudocounts)"%(sys.argv[1],total)
for nmer in _t:
print "%-7s %20.17f"%(nmer,float(nmersD[nmer]) / total)
sys.stdout.flush()
def main(fastafile, outDirectory): # !! 1/2/09 AD added 'fastafile' var and changed 'if __name__' as way to call this from script.
seqsD = Fasta.load(fastafile)
seqs = seqsD.values()
output = []
for w in range(1,7):
allnmers = permute(w)
nmersT = MotifTools.top_nmers(w,seqs,'with counts','purge Ns')
nmersD = {}
total = 0
for nmer in allnmers:
nmersD[nmer] = 1 #Pseudo count
total = total + 1
for nmer,count in nmersT[:]:
try:
rc = MotifTools.revcomplement(nmer)
nmersD[nmer] = nmersD[nmer] + count
nmersD[rc] = nmersD[rc] + count
total = total + 2*count
except KeyError:
pass
_t = nmersD.keys()
_t.sort()
output.append("# freq in %s (total %d with pseudocounts)\n"%(fastafile.split('/')[-1],total)) # AD 02-27-09 added a '\n' to make file look right
for nmer in _t:
output.append( "%-7s %20.17f\n"%(nmer,float(nmersD[nmer]) / total)) # AD 02-27-09 added a '\n' to make file look right
# open output file and write out results
outFile = '%s/%s.freq' % (outDirectory, fastafile.split('/')[-1])
outFile = open(outFile, 'w')
for index in output:
outFile.write(index)
def study_seqs(self,seqs):
for depth in range(1,6):
nmersT = MotifTools.top_nmers(depth, seqs, "TUPLES")
total = 0
for nmer,count in nmersT:
total = total + count
rc = MotifTools.revcomplement(nmer)
for nmer,count in nmersT:
f = math.log(float(count)/total)/math.log(2)
f_2 = math.log(0.5 * float(count)/total)/math.log(2)
rc = MotifTools.revcomplement(nmer)
if rc != nmer:
self.D[nmer] = f_2
self.D[rc] = f_2
else:
self.D[nmer] = f
for depth in range(0):
total = 0
for k in self.D.keys():
if len(k) == depth:
total = total + pow(2,self.D[k])
print k, pow(2,self.D[k])
print depth,total
self.highestorder = 5
def freq_from_seqs(self,seqs):
self.highestorder = 6
for w in range(1,7):
allnmers = permute(w)
nmersT = MotifTools.top_nmers(w,seqs,'with counts','purge Ns')
self.nmers_by_size[w] = allnmers[:]
nmersD = {}
total = 0.0
for nmer in allnmers: #Pseudo count
nmersD[nmer] = 1
total = total + 1
for nmer,count in nmersT:
try:
rc = MotifTools.revcomplement(nmer)
nmersD[nmer] = nmersD[nmer] + count
nmersD[rc] = nmersD[rc] + count
total = total + 2*count
except KeyError:
pass
for nmer in nmersD.keys():
rc = MotifTools.revcomplement(nmer)
f = nmersD[nmer]/total
self.F[nmer] = f
self.F[rc] = f
def freq_from_seqs_old(self,seqs):
self.highestorder = 4
for depth in range(1,6):
nmersT = MotifTools.top_nmers(depth, seqs, "TUPLES")
self.nmers_by_size[depth] = map(lambda x:x[0],nmersT)
total = 0
for nmer,count in nmersT:
total = total + count
for nmer,count in nmersT:
rc = MotifTools.revcomplement(nmer)
if nmer == rc: #correct top_nmers
f = float(count)/total #palindrome count
else:
f = float(count)/total/2
self.F[nmer] = f
self.F[rc] = f
for depth in range(0): #For debugging
total = 0
for k in self.F.keys():
if len(k) == depth:
total = total + self.F[k]
print k, self.F[k]
print depth,total
def all_Wmers(self,N,seq):
forw = []
rev = []
seqrc = MotifTools.revcomplement(seq)
Mlh = theMarkovBackground.highestorder
Mlb = theMarkovBackground.logbackground
MCP = theMarkovBackground.CP
Fbg = Mlb(seq)
Rbg = Mlb(seqrc)
nmask = map(lambda x:1-x, self.mask)
'''
?? QUESTION: Is it sensible to compute the background probabilities
this way?
1) BG of complementary strand is taken as equal to primary strand.
2) Letters inside the motif window are not used for conditional probabilities.
As a result, the calculation essentially breaks down to the log probability the
background emits the sequence to the left of the window plus the log probability
the background emits the sequence to the right.
3) I\'ve worked out an efficient way to compute this by
a) Compute the background probability for the entire probe/sequence
b) (Quick) Compute logQdiff below
c) Subtract
'''
for i in range(len(seq)-N+1):
subseq = seq[i:i+N]
'''Build Wmer information'''
#Wtmp = Wmer(subseq)
left = seq[0:i]
right = seq[i+N:]
#Wtmp.lflank = left
#Wtmp.rflank = right
#if i==0: Wtmp.src = seq
#Wtmp.srcQ = Fbg
#Wtmp.i = i
'''This is the fast way'''
logQdiff = Mlb(left[-Mlh:] + subseq + right[0:Mlh]) - Mlb(left[-Mlh:]) - Mlb(right[0:Mlh])
logQtot = Fbg - logQdiff
'''Add a bit back for intervening bases in the "gap" '''
gapbg = 0
for p in range(N):
gapbg = gapbg + MCP[subseq[p]] * nmask[p]
logQtot = logQtot + gapbg
'''Build Wmer-reverse complement information'''
#Wtmprc = Wmer(Wtmp.rc)
#Wtmprc.lflank = seqrc[0:-(i+N)] #Check this in case it is ever necessary
#if i!=0:
# Wtmprc.rflank = seqrc[-i:] #Necessary [11-12-02]
#else:
# Wtmprc.rflank = ''
#Wtmprc.logQtot = Wtmp.logQtot
#Wtmprc.srcQ = Wtmp.srcQ
#Wtmprc.i = i
forw.append(logQtot)
rev.append(logQtot)
W = []
W.extend(forw)
W.extend(rev)
#seq.c_wmerbgs = MDsupport.list2double(map(lambda x: x.logQtot, W))
#MDsupport.printdouble(seq.c_wmerbgs,len(W))
return(W)
def has_wmer(self,wmer):
rc = MotifTools.revcomplement(wmer)
if (wmer in self.wmers) or (rc in self.wmers):
return(1)
else:
return(0)
def probOvlp(A,B,thresh=0.7,verbose=None):
if A.width >= B.width:
Wide, Narrow = A, B
else:
Wide, Narrow = B, A
RC = MotifTools.revcomplement
if 1:
newWide = Wide[-1,Wide.width+1]
if Wide.__dict__.has_key('bestWide'):
bestWide = Wide.bestWide
else:
bestWideD = {}
for x in newWide.bestseqs(thresh*newWide.maxscore):
bestWideD[x] = 1
for x in bestWideD.keys():
bestWideD[RC(x)] = 1
Wide.bestWide = bestWideD.keys()
bestWide = Wide.bestWide
Wide = newWide
if Narrow.__dict__.has_key('bestNarrow'):
bestNarrow = Narrow.bestNarrow
else:
bestNarrowD = {}
for x in Narrow.bestseqs(thresh*Narrow.maxscore):
bestNarrowD[x] = 1
for x in bestNarrowD.keys():
bestNarrowD[RC(x)] = 1
bestNarrow = bestNarrowD.keys()
Narrow.bestNarrow = bestNarrow
#bestWide = [x[1] for x in Wide.bestseqs (thresh*Wide.maxscore) ]
#bestNarrow = [x[1] for x in Narrow.bestseqs(thresh*Narrow.maxscore)]
countNarrow = len(bestNarrow)
countWide = len(bestWide)
numtotal = math.pow(4,Wide.width)
fudgefactor = math.pow(4,Wide.width - Narrow.width)
bestWideTups = [(x,MotifTools.revcomplement(x)) for x in bestWide]
countBoth = 0
for i in range(len(bestNarrow)):
m_narrow = bestNarrow[i]
delj = []
for j in range(len(bestWideTups)):
if (bestWideTups[j][0].find(m_narrow) >= 0) or (bestWideTups[j][1].find(m_narrow) >= 0):
countBoth += 1
delj.append(j)
delj.reverse() #Chew in from the back
for j in delj:
del(bestWideTups[j])
if verbose: print '%10d %10d %10d %10d | %10d %5d '%(
countWide, numtotal, countNarrow *fudgefactor , countBoth , countNarrow, Wide.width - Narrow.width),
p = Arith.hypgeomsummore(countWide, #Num Interesting
numtotal, #All k-mers
countNarrow * fudgefactor, #Number picked
countBoth ) #Number found
return p
def probOvlp(A, B, thresh=0.7, verbose=None):
if A.width >= B.width:
Wide, Narrow = A, B
else:
Wide, Narrow = B, A
RC = MotifTools.revcomplement
if 1:
newWide = Wide[-1, Wide.width + 1]
if Wide.__dict__.has_key('bestWide'):
bestWide = Wide.bestWide
else:
bestWideD = {}
for x in newWide.bestseqs(thresh * newWide.maxscore):
bestWideD[x] = 1
for x in bestWideD.keys():
bestWideD[RC(x)] = 1
Wide.bestWide = bestWideD.keys()
bestWide = Wide.bestWide
Wide = newWide
if Narrow.__dict__.has_key('bestNarrow'):
bestNarrow = Narrow.bestNarrow
else:
bestNarrowD = {}
for x in Narrow.bestseqs(thresh * Narrow.maxscore):
bestNarrowD[x] = 1
for x in bestNarrowD.keys():
bestNarrowD[RC(x)] = 1
bestNarrow = bestNarrowD.keys()
Narrow.bestNarrow = bestNarrow
#bestWide = [x[1] for x in Wide.bestseqs (thresh*Wide.maxscore) ]
#bestNarrow = [x[1] for x in Narrow.bestseqs(thresh*Narrow.maxscore)]
countNarrow = len(bestNarrow)
countWide = len(bestWide)
numtotal = math.pow(4, Wide.width)
fudgefactor = math.pow(4, Wide.width - Narrow.width)
bestWideTups = [(x, MotifTools.revcomplement(x)) for x in bestWide]
countBoth = 0
for i in range(len(bestNarrow)):
m_narrow = bestNarrow[i]
delj = []
for j in range(len(bestWideTups)):
if (bestWideTups[j][0].find(m_narrow) >=
0) or (bestWideTups[j][1].find(m_narrow) >= 0):
countBoth += 1
delj.append(j)
delj.reverse() #Chew in from the back
for j in delj:
del (bestWideTups[j])
if verbose:
print '%10d %10d %10d %10d | %10d %5d ' % (
countWide, numtotal, countNarrow * fudgefactor, countBoth,
countNarrow, Wide.width - Narrow.width),
p = Arith.hypgeomsummore(
countWide, #Num Interesting
numtotal, #All k-mers
countNarrow * fudgefactor, #Number picked
countBoth) #Number found
return p
