def computeExpectedCodonProbForSeqs(seqDict,
rf,
gcDiff,
constraintfile=None,
filename=None,
threshold=1):
#compute probability of codon w for a set of nucleotide sequences
totalfreq = {}
totalfreqSyn = {}
for codon in genCode.keys():
totalfreq[codon] = 0.0
totalfreqSyn[codon] = 0.0
p_hyp = {}
totCDS = 0
for codon in genCode.keys():
p_hyp[codon] = 0.0
for ids, seq in seqDict.iteritems():
print ids
peptides = {}
if (constraintfile):
constraint = parseConstraintFile(constraintfile, len(seq))
else:
constraint = None
#---------------generate peptides for the given rf ------------
#translate sequences
for r in rf:
if r[0] == '+':
peptides[r] = translateMRNA(seq[int(r):])
elif r[0] == '-':
peptides[r] = translateMRNA(seq[::-1][int(r) + 2:])
n = len(peptides[r])
#if the length of the peptides in all reading frames are not identical make them equal
N = []
N.append(n)
for pep in peptides.values():
if len(pep) != n:
N.append(len(pep))
n = min(N)
if len(N) != 1:
for k in peptides.keys():
peptides[k] = peptides[k][:n]
#add the 'O' peptides in reading frames that are not defined in the input
for i in ['+0', '+1', '+2', '-0', '-1', '-2']:
if i not in peptides.keys():
peptides[i] = 'O' * n
if DEBUG: print peptides
print peptides
print constraint
#--------------compute pf and frequency dictionary-------------
if gcDiff == -1: #no gc-content defined
ZF = computeZF(peptides, constraint, filename, threshold)
ZB = computeZB(peptides, constraint, filename, threshold)
ZFval = 0.0
for ch3 in NUCL:
for ch4 in NUCL:
ZFval += ZF[(n, ch3, ch4)]
print ZFval
totCDS += ZFval
gclow = -1
gcup = -1
else: #gc-content defined
ZF = computeZF_GC(peptides, constraint, filename, threshold)
ZB = computeZB_GC(peptides, constraint, filename, threshold)
cnt = round((gcDiff / 100.0) * len(seq))
gc = GCcont(seq)
gclow = gc - cnt
gcup = gc + cnt
if (gclow < 0): gclow = 0
if (gcup > 3 * n + 2): gcup = 3 * n + 2
#~ print len(seq),gc,gclow,gcup
#~ ZBval_gc = 0.0
#~ for ch1 in NUCL:
#~ for ch2 in NUCL:
#~ for x in range(3*n+3):
#~ ZBval_gc += ZF[(n,x,ch1,ch2)]
#~ #print "gc,len",gc,len(seq)
#~ print "total:",ZBval_gc
#~
#~ ZBval = 0.0
#~ for ch1 in NUCL:
#~ for ch2 in NUCL:
#~ for x in range(3*n+3):
#~ ZBval += ZB[(0,x,ch1,ch2)]
#~ print "totalZB:",ZBval
#~
ZFval = 0.0
for ch3 in NUCL:
for ch4 in NUCL:
for x in range(int(gclow), int(gcup + 1)):
ZFval += ZF[(n, x, ch3, ch4)]
#print ZFval
totCDS += ZFval
freq, freq_syn = computeExpectedCodonFreqForPeptide(
rf, peptides, ZF, ZB, gclow, gcup, filename, threshold)
for codon in genCode.keys():
totalfreq[codon] += freq[codon]
totalfreqSyn[codon] += freq_syn[codon]
print "#total number of sequences that code exactly the same peptides as the given fasta file: ", totCDS
for codon in genCode.keys():
if totalfreqSyn[codon] != 0:
p_hyp[codon] = totalfreq[codon] / totalfreqSyn[codon]
return p_hyp, totalfreq, totalfreqSyn
def checkCodonUsage(seqList, rf, gcDiff, constraint=None):
#codon usage computed directly from the sequences must be identical to the codon usage computed from partition functions
l = 0
RNA0 = []
for seq in seqList:
if (constraintfile):
constraint = parseConstraintFile(constraintfile, len(seq))
else:
constraint = None
p1 = {}
for codon in genCode.keys():
p1[codon] = 0.0
peptides = {}
#translate sequences
for r in rf:
if r[0] == '+':
peptides[r] = translateMRNA(seq[int(r):])
elif r[0] == '-':
peptides[r] = translateMRNA(seq[::-1][int(r) + 2:])
n = len(peptides[r])
#if the length of the peptides in all reading frames are not identical make them equal
N = []
N.append(n)
for pep in peptides.values():
if len(pep) != n:
N.append(len(pep))
n = min(N)
if len(N) != 1:
for k in peptides.keys():
peptides[k] = peptides[k][:n]
#add the 'O' peptides in reading frames that are not defined in the input
for i in ['+0', '+1', '+2', '-0', '-1', '-2']:
if i not in peptides.keys():
peptides[i] = 'O' * n
if DEBUG: print peptides
#print peptides
RNA = bfs(peptides, filename, threshold, constraint)
if gcDiff == -1:
for rna in RNA:
RNA0.append(rna)
#~ for rna in RNA:
#~ print rna,GCcont(rna)
else: #gc-content defined
cnt = round((gcDiff / 100.0) * len(seq))
gc = GCcont(seq)
gclow = gc - cnt
gcup = gc + cnt
if (gclow < 0): gclow = 0
if (gcup > 3 * n + 2): gcup = 3 * n + 2
print gc, gclow, gcup
seqLen = len(seqList[0])
for rna in RNA:
if int(gclow <= GCcont(rna) <= gcup):
print rna, GCcont(rna)
RNA0.append(rna)
l = len(RNA0)
p1, f1, f2 = computeCodonProbFromGenesOverlap(RNA0, rf)
print '#number of seq from bruteforce: %d' % l
p2, f3, f4 = computeExpectedCodonProbForSeqs(seqList, rf, gcDiff)
for codon in genCode.keys():
print codon, p1[codon], p2[codon]
assert (p1 == p2)
def computeCodonProbFromGenesOverlap(seqList, rf):
#compute probability of codon w in the input sequence list 'seqList' for the overlapping reading frames given in rf. For each reading frame of rf
#compute the freq(w)/freq(syn(w)) in the corresponding reading frame.
#Output is dictionary p_gene where keys are codons and values are sum of the probabilities over all reading frames.
#-------------------initialize the codon frequency dictionary
freq = {}
for cod in genCode.keys():
freq[cod] = 0.0
#------------------check the length of the sequences to be 3k+2-----------------------------------
#~ m = n%3
#~ l = []
#~ if m!=2:
#~ for seq in seqList:
#~ l.append(seq[:n-m-1])
#~ print "WARNING: the last %d nucleotides ignored in computing codon usage from genes!" % (m+1)
#~ seqList = l
#~ if l==[]:
#~ print 'error: sequences are to short!'
#~ sys.exit(1)
#~ n = len(seqList[0])
#------------------calculate frequency of codons in all reading frames --------------------------------
for i in rf:
for seq in seqList:
n = len(seq)
if i[0] == '+': #positive strands
r = int(i)
for idx in range(
r, n - 2 + r, 3
): #seq[0:n-2], seq[1:n-1], seq[2,n] code in reading frames +0, +1 and +2 respectively
codon = seq[idx:idx + 3]
if len(codon) == 3:
freq[codon] += 1
elif i[0] == '-': #negative strands
r = abs(int(i))
for idx in range(
(n - 2 + r), r, -3
): #seq[0:n-2], seq[1:n-1], seq[2,n] code in reading frames +0, +1 and +2 respectively
codon = seq[idx - 3:idx][::-1]
if len(codon) == 3:
freq[codon] += 1
if DEBUG:
print "codon freq in the fasta file:"
for k in sorted(freq.keys()):
if freq[k] != 0:
print k, freq[k]
#----------------calculate frequency of synonymous codons in all reading frames-------------------------------
freq_syn = {}
for codon in genCode.keys():
freq_syn[codon] = 0.0
for codon in freq.keys():
for syncod in aa2codon[aaCode[genCode[codon].upper()]]:
freq_syn[codon] += freq[syncod]
if DEBUG:
print "total synonymous codon freq in the fasta file: "
for k in sorted(freq_syn.keys()):
if freq_syn[k] != 0:
print k, aa2codon[aaCode[
genCode[k].upper()]], freq[k], freq_syn[k]
#---------------calculate the codon usage index-----------------------------------------------------------------
p = {}
for codon in genCode:
p[codon] = 0.0
for codon in freq.keys():
if freq_syn[codon] != 0:
p[codon] = freq[codon] / freq_syn[codon]
return p, freq, freq_syn
def computeExpectedCodonFreqForPeptide(rf,
peptides,
ZF,
ZB,
gclow,
gcup,
filename=None,
threshold=1):
#compute probability of codon w in the set of all sequences that translate 'peptides' in the corresponding reading frames
n = len(peptides['+0'])
seqLen = 3 * n + 2
#----------Initialize freq ---------------
freq = {}
for codon in genCode.keys():
freq[codon] = 0.0
#-----------calculate frequency of codons in all reading frames of interest -------------------
ZFval = 0.0
if gclow != -1: #if gc-content is defined
for ch3 in NUCL:
for ch4 in NUCL:
for x in range(int(gclow), int(gcup + 1)):
ZFval += ZF[(n, x, ch3, ch4)]
for k in range(n):
hexa = [
peptides['+0'][k], peptides['+1'][k], peptides['+2'][k],
peptides['-0'][n - k - 1], peptides['-1'][n - k - 1],
peptides['-2'][n - k - 1]
]
L = createListOfCompatible5mers(hexa, filename, threshold)
for s in L:
for x1 in range(0, 3 * k + 3):
for x2 in range(0, 3 * (n - k) + 3):
#print s,k,x1,x2,ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])]
if (gclow <= x1 + x2 + GCcont(s[2]) <= gcup):
#print '*',k,s,GCcont(s[2]),x1,x2, ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])]
if '+0' in rf:
freq[s[0:3]] += (ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3],
s[4])]
if '+1' in rf:
freq[s[1:4]] += (ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3],
s[4])]
if '+2' in rf:
freq[s[2:5]] += (ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3],
s[4])]
if '-0' in rf:
freq[s[0:3][::-1]] += (
ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3], s[4])]
if '-1' in rf:
freq[s[1:4][::-1]] += (
ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3], s[4])]
if '-2' in rf:
freq[s[2:5][::-1]] += (
ZF[(k, x1, s[0], s[1])] /
ZFval) * ZB[(k + 1, x2, s[3], s[4])]
#~ if ZF[(k,x1,s[0],s[1])] * ZB[(k+1,x2,s[3],s[4])]!= 0:
#~ print 's:%s\tgc2:%d\tk:%d\tx1:%d\tx2:%d\tZF[k,x1,s[0],s[1]]:%f\tZB[(k+1,x2,s[3],s[4])]:%d\tfreq:%d'%(s,GCcont(s[2]),k,x1,x2, ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])],ZF[(k,x1,s[0],s[1])]*ZB[(k+1,x2,s[3],s[4])])
else: #if gc-content is not defined
for ch3 in NUCL:
for ch4 in NUCL:
ZFval += ZF[(n, ch3, ch4)]
for k in range(n):
hexa = [
peptides['+0'][k], peptides['+1'][k], peptides['+2'][k],
peptides['-0'][n - k - 1], peptides['-1'][n - k - 1],
peptides['-2'][n - k - 1]
]
L = createListOfCompatible5mers(hexa, filename, threshold)
for s in L:
if '+0' in rf:
freq[s[0:3]] += (ZF[(k, s[0], s[1])] /
ZFval) * ZB[(k + 1, s[3], s[4])]
if '+1' in rf:
freq[s[1:4]] += (ZF[(k, s[0], s[1])] /
ZFval) * ZB[(k + 1, s[3], s[4])]
if '+2' in rf:
freq[s[2:5]] += (ZF[(k, s[0], s[1])] /
ZFval) * ZB[(k + 1, s[3], s[4])]
if '-0' in rf:
freq[s[0:3][::-1]] += (ZF[(k, s[0], s[1] / ZFval)]) * ZB[
(k + 1, s[3], s[4])]
if '-1' in rf:
freq[s[1:4][::-1]] += (ZF[(k, s[0], s[1] / ZFval)]) * ZB[
(k + 1, s[3], s[4])]
if '-2' in rf:
freq[s[2:5][::-1]] += (ZF[(k, s[0], s[1] / ZFval)]) * ZB[
(k + 1, s[3], s[4])]
#~ ###
#~ if gclow != -1: #if gc-content is defined
#~ for k in range(n):
#~ hexa = [peptides['+0'][k], peptides['+1'][k], peptides['+2'][k], peptides['-0'][n-k-1],peptides['-1'][n-k-1],peptides['-2'][n-k-1]]
#~ L = createListOfCompatible5mers(hexa,filename,threshold)
#~ for s in L:
#~ for x1 in range(0,3*k+3):
#~ for x2 in range(0,3*(n-k)+3):
#~ #print s,k,x1,x2,ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])]
#~ if( gclow <= x1 + x2 + GCcont(s[2]) <= gcup) :
#~ #print '*',k,s,GCcont(s[2]),x1,x2, ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])]
#~ if '+0' in rf : freq[s[0:3]]+= (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#~ if '+1' in rf : freq[s[1:4]]+= (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#~ if '+2' in rf : freq[s[2:5]] += (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#~ if '-0' in rf : freq[s[0:3][::-1]]+= (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#~ if '-1' in rf : freq[s[1:4][::-1]]+= (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#~ if '-2' in rf : freq[s[2:5][::-1]] += (ZF[(k,x1,s[0],s[1])]) * ZB[(k+1,x2,s[3],s[4])]
#if ZF[(k,x1,s[0],s[1])] * ZB[(k+1,x2,s[3],s[4])]!= 0:
#print 's:%s\tgc2:%d\tk:%d\tx1:%d\tx2:%d\tZF[k,x1,s[0],s[1]]:%f\tZB[(k+1,x2,s[3],s[4])]:%d\tfreq:%d'%(s,GCcont(s[2]),k,x1,x2, ZF[(k,x1,s[0],s[1])],ZB[(k+1,x2,s[3],s[4])],ZF[(k,x1,s[0],s[1])]*ZB[(k+1,x2,s[3],s[4])])
#~ else: #if gc-content is not defined
#~ for k in range(n):
#~ hexa = [peptides['+0'][k], peptides['+1'][k], peptides['+2'][k], peptides['-0'][n-k-1],peptides['-1'][n-k-1],peptides['-2'][n-k-1]]
#~ L = createListOfCompatible5mers(hexa,filename,threshold)
#~ for s in L:
#~ if '+0' in rf : freq[s[0:3]]+= (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~ if '+1' in rf : freq[s[1:4]]+= (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~ if '+2' in rf : freq[s[2:5]] += (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~ if '-0' in rf : freq[s[0:3][::-1]]+= (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~ if '-1' in rf : freq[s[1:4][::-1]]+= (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~ if '-2' in rf : freq[s[2:5][::-1]] += (ZF[(k,s[0],s[1])]) * ZB[(k+1,s[3],s[4])]
#~
###
#~ if ZF[(k,s[0],s[1])] * ZB[(k+1,s[3],s[4])]!= 0:
#~ print k,s,ZF[(k,s[0],s[1])],ZB[(k+1,s[3],s[4])],freq[s[0:3]]
#----------check frequency matrix (only works without blosum)--------------------------------------------------------
sum0 = 0.0
ZFval = 0.0
for k in freq.keys():
sum0 += freq[k]
if gclow != -1:
for ch3 in NUCL:
for ch4 in NUCL:
for x in range(int(gclow), int(gcup) + 1):
ZFval += ZF[(n, x, ch3, ch4)]
print "####", ZFval
else:
for ch1 in NUCL:
for ch2 in NUCL:
ZFval += ZF[(n, ch1, ch2)]
print(sum0, ZFval * len(rf) * n)
#assert(sum0 == ZFval*len(rf)*n)
#~ #print (sum0 , ZFval*len(rf)*n)
#----------calculate frequency of synonymous codons-----------------------------------------
freq_syn = {}
for codon in genCode.keys():
freq_syn[codon] = 0.0
for codon in freq.keys():
for syncod in aa2codon[aaCode[genCode[codon].upper()]]:
freq_syn[codon] += freq[syncod]
if DEBUG:
print 'expected codon freq for peptides:', peptides
for k in sorted(freq_syn.keys()):
if freq_syn[k] != 0:
#print k,aa2codon[aaCode[genCode[k].upper()]],freq[k],freq_syn[k]
print k, freq[k], freq_syn[k]
#---------------calculate the codon usage index-----------------------------------------------------------------
#~ p = {}
#~ for codon in genCode.keys():
#~ p[codon]=0.0
#~ for codon in freq.keys():
#~ if freq_syn[codon]!=0:
#~ assert((freq[codon]/freq_syn[codon])<=1)
return freq, freq_syn
#GC content with -gc flag
for i in range(len(args)):
if args[i] == '-gc':
gcDiff = int(args[i + 1])
del [args[i + 1]]
del [args[i]]
break
#verbose output, with -v flag
for i in range(len(args)):
if args[i] == '-v':
VERBOSE = 1
del [args[i]]
break
if rf == []: #the above check does not work when only -f is defined
print "coding reading frames must be specified!\n"
printusage(sys.argv[0])
seqDict = parseFasta(fastafile)
#checkCodonUsage(seqList,rf,gcDiff,constraintfile)
p_gene, f_gene, fsyn_gene = computeCodonProbFromGenesOverlap(
seqDict.values(), rf)
p_hyp, f_hyp, fsyn_hyp = computeExpectedCodonProbForSeqs(
seqDict, rf, gcDiff, constraintfile, filename, threshold)
for codon in genCode.keys():
if p_hyp[codon] != 0:
print codon, f_gene[codon], fsyn_gene[codon], f_hyp[
codon], fsyn_hyp[codon], p_gene[codon], p_hyp[
codon], p_gene[codon] / p_hyp[codon]
else:
print codon, f_gene[codon], fsyn_gene[codon], f_hyp[
codon], fsyn_hyp[codon], p_gene[codon], p_hyp[codon], 'NA'