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 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)