def relEntropy(motif1, motif2):
output = 0.0
try:
for i in range(0, len(motif1)):
for j in range(0,4):
try:
output = output + motif1[i][j] * math.log(motif1[i][j] / motif2[i][j])
except ZeroDivisionError:
output += 0
except IndexError:
output = -100.54
return output
if __name__ == "__main__":
print "hello evaluation"
predMotifFiles = directory.getFiles('predictedmotif.txt') #array of file names
motifFiles = directory.getFiles('motif.txt')
entropy_result = computeEntropy(motifFiles, predMotifFiles)
fo = open('entropy_output.txt','w')
for x in entropy_result:
fo.write("" +x[0]+','+str(x[1]) +"\n")
fo.close()
sitesFiles = directory.getFiles('sites.txt')
predSitesFiles = directory.getFiles('predictedsites.txt')
site_result = siteOverlap(sitesFiles,predSitesFiles)
fo = open('site_output.txt','w')
for x in site_result:
fo.write("" +x[0]+','+str(x[1]) +"\n")
def chooseMotifPositions():
positions = []
for x in range(0, globals.numberOfSequences):
positions.append(random.randint(0, globals.lengthOfSequences - globals.motifLength))
return positions
if __name__ == "__main__":
print "\nRunning motif finder..."
print "-----------------------" + '\n'
startTime = datetime.datetime.now()
print "start time = " + str(startTime) + '\n'
# create array of motif length text files
motifLengthFiles = directory.getFiles('motiflength.txt')
sequencesFiles = directory.getFiles('sequences.fa')
numberOfFiles = len(motifLengthFiles)
# iterate over arrays
informationContent = 0
for x in range(0, numberOfFiles):
sequencesFile = sequencesFiles[x]
motifLengthFile = motifLengthFiles[x]
sequences = reader.readFastaFile(sequencesFile)
motifLength = reader.readMotifLengthFile(motifLengthFile)
output = findMotif(sequences, motifLength, int(sys.argv[1]))
informationContent += output[2]
writer.writePredictions(output, sequencesFile, motifLength)
print "files written for " + str(sequencesFile)
if (x + 1) % 10 == 0:
