def toTAMOmotifs(motifList,seqData=None):
"""Returns a list of TAMO motif Objects when given
a list of the native motif Objs.
If seqData is a dict, it is assumed to be
nucFreqs (seqData['A']==.26),
else, it is assumed to be a path to a seqPopulation
that the motifs will be used on for stats to be
calculated against."""
from TAMO.MotifTools import Motif_from_counts
if not seqData:
if type(seqData) == type({}):
try:
aFreq = seqData['A']
cFreq = seqData['C']
gFreq = seqData['G']
tFreq = seqData['T']
except KeyError:
raise InvalidOptionError("toTAMOmotifs: Unrecognized key in seqData as dict.")
elif type(seqData) == type(''):
sDict = ParseFastA(seqData).toDict()
seqData = seqStats(sDict)
del(sDict)
tot = float(seqData['nonNs'])
aFreq = seqData['aCnt']/tot
cFreq = seqData['cCnt']/tot
gFreq = seqData['gCnt']/tot
tFreq = seqData['tCnt']/tot
else:
aFreq = 0.25
cFreq = 0.25
gFreq = 0.25
tFreq = 0.25
tList = []
for m in motifList:
counts = []
for i in range(len(m)):
counts.append({'A':m.pwm['A'][i],'C':m.pwm['C'][i],'G':m.pwm['G'][i],'T':m.pwm['T'][i]})
t = Motif_from_counts(counts,beta=0.01,bg={'A':aFreq,'C':cFreq,'G':gFreq,'T':tFreq})
t.id = m.id
try:
t.sigvalue = m.sigvalue
except AttributeError:
pass
tList.append(t)
return tList
tamoMotifs = []
for jMat in jasparFiles:
tempMat = map(lambda l: l.strip().split(), open(jMat, "rU").readlines())
## eval() inteligently converts text numbers to int or float!
# for i in range(len(tempMat)):
# for j in range(len(tempMat[i])):
# tempMat[i][j] = eval(tempMat[i][j])
# transpose matrix
tempMat = transpose(mat(tempMat)).tolist()
for i in range(len(tempMat)):
tempMat[i] = {
"A": eval(tempMat[i][0]),
"C": eval(tempMat[i][1]),
"G": eval(tempMat[i][2]),
"T": eval(tempMat[i][3]),
}
jasparTAMO_Motif = Motif_from_counts(tempMat, bg=nucBack)
jasparTAMO_Motif.sourceFile = jMat.split("/")[-1]
tamoMotifs.append(jasparTAMO_Motif)
# print to file
cPickle.dump(tamoMotifs, open(pklFile, "w"))
save_motifs(tamoMotifs, tmoFile, kmer_count=60)
print "Done."
