def mainRunner(inputFileName, initialProbabilities, transitionProbalities, emissionI, emissionD, converstionTable, delim, initial):
try:
inputFile = open(inputFileName, 'r')
except IOError:
print 'The input file does not exist to read'
exit(1)
sequences = stripAwayNewLines(inputFile) #tuple pair of each sequence
seqLength = len(sequences[0])
inputFile.close()
print "Calculating Forward"
forwardTableLog = calculateForwardAlgoLog(initialProbabilities, transitionProbalities, emissionI, emissionD, sequences )
print "Calculatring Probability of Path"
likelihood = likelihoodOfSequence(forwardTableLog)
print "Calculating Backwards"
backwardsTableLog = calculateBackwardsAlgoLog(transitionProbalities, emissionI, emissionD, sequences )
print "Calculating Viterbi Encoding"
veterbiEncodingLog = calculateVeterbiEncodingLog(initialProbabilities, transitionProbalities, emissionI, emissionD, sequences )
veterbiStatesLog = hiddenStatePath(veterbiEncodingLog[1], seqLength, getLastState(veterbiEncodingLog[0],seqLength ))
print "Calculating Most Probable States Posterior Encoding"
posteriorTableAndRouteLog = computePosteriorDecodingLog(forwardTableLog, backwardsTableLog, seqLength, likelihood)# tuple pair. Has all post values and most probable
posteriorStatesLog = posteriorTableAndRouteLog[1]
posteriorMeanLog = calculatePosteriorMeanLog(posteriorTableAndRouteLog[0], seqLength, converstionTable)
outputFile = open(delim +"/"+initial+"_veterbiEncoding_" + delim + ".txt", 'w')
writeToFileWithBreaks(outputFile, veterbiStatesLog, converstionTable)
print "Finished Writing Veterb Encoding for " + delim
outputFile.close()
outputFile2 = open(delim +"/"+initial+"_Posterior_"+ delim + ".txt", 'w')
writeToFileWithBreaks(outputFile2, posteriorStatesLog, converstionTable)
print "Finished Writing Posterior Encoding for " + delim
outputFile2.close()
outputFile3 = open(delim +"/"+initial+"_PosteriorMean_"+delim + ".txt", 'w')
writeToFileWithBreaksStraight(outputFile3, posteriorMeanLog)
print "Finished Writing Posterior Mean for " + delim
outputFile3.close()
outPutFile4 = open(delim +"/"+initial+"_ThreeCol_"+delim + ".txt", 'w')
writeToFileThreeCol(outPutFile4, (veterbiStatesLog,posteriorStatesLog, posteriorMeanLog ), converstionTable)
print "Finished Writing Three Columns for " + delim
outPutFile4.close()
outPutFile5 = open(delim +"/"+initial+"_likelihoods_"+delim + ".txt", 'w')
outPutFile5.write(str(likelihood) + '\n')
print "Finished Writing Log-likelihoods for " + delim
outPutFile5.close()
print "Done"
def expectation_maximization(sequences, params):
forward_table = calculateForwardAlgoLog(params['m'], params['t'], params['e']['i'], params['e']['d'], sequences)
likelihood = likelihoodOfSequence(forward_table)
backward_table = calculateBackwardsAlgoLog(params['t'], params['e']['i'], params['e']['d'], sequences)
posterior_table_and_route_log = computePosteriorDecodingLog(forward_table, backward_table, len(sequences[0]), likelihood)
check = calculateBigVars (posterior_table_and_route_log[0], sequences, forward_table, backward_table, params['t'], params['e']['i'], params['e']['d'], likelihood)
results = calculateNewInitialValues(check)
params['m'] = results[0]
params['t'] = results[1]
params['e']['i'] = results[2]
params['e']['d'] = results[3]
return params, likelihood
def viterbi(sequences, params):
decoding = []
forward_table = calculateForwardAlgoLog(params['m'], params['t'], params['e']['i'], params['e']['d'], sequences)
likelihood = likelihoodOfSequence(forward_table)
backward_table = calculateBackwardsAlgoLog(params['t'], params['e']['i'], params['e']['d'], sequences)
viterbi_encoding_log = calculateVeterbiEncodingLog(params['m'], params['t'], params['e']['i'], params['e']['d'], sequences)
viterbi_states_log = hiddenStatePath(viterbi_encoding_log[1], len(sequences[0]), getLastState(viterbi_encoding_log[0], len(sequences[0])))
posterior_table_and_route_log = computePosteriorDecodingLog(forward_table, backward_table, len(sequences[0]), likelihood)
posterior_states_log = posterior_table_and_route_log[1]
posterior_mean_log = calculatePosteriorMeanLog(posterior_table_and_route_log[0], len(sequences[0]), conversion_table)
decoding.extend('# Viterbi_decoding posterior_decoding posterior_mean\n')
for i in range(len(viterbi_states_log)):
decoding.extend('{0} {1} {2}\n'.format(conversion_table[viterbi_states_log[i]], conversion_table[posterior_states_log[i]], posterior_mean_log[i]))
return likelihood, ''.join(decoding)
def mainRunnerEM(inputFileName, initialProbabilities, transitionProbalities, emissionI, emissionD, converstionTable, delim):
try:
inputFile = open(inputFileName, 'r')
except IOError:
print 'The input file does not exist to read'
exit(1)
sequences = stripAwayNewLines(inputFile) #tuple pair of each sequence
seqLength = len(sequences[0])
inputFile.close()
forwardTable = calculateForwardAlgoLog(initialProbabilities, transitionProbalities, emissionI, emissionD, sequences )
likelihood = likelihoodOfSequence(forwardTable)
backwardTable = calculateBackwardsAlgoLog(transitionProbalities, emissionI, emissionD, sequences )
posteriorTableAndRouteLog = computePosteriorDecodingLog(forwardTable, backwardTable, seqLength, likelihood)# tuple pair. Has all post values and most probable
check = calculateBigVars (posteriorTableAndRouteLog[0], sequences, forwardTable, backwardTable, transitionProbalities, emissionI, emissionD, likelihood) #E
temp = calculateNewInitialValues(check) #M
return (temp, likelihood)