def main():
mutationmatrix = '/Users/jlu96/maf/new/PRAD_broad/PRAD_broad-som.m2'
patientFile = None #'/Users/jlu96/maf/new/PRAD_broad/shared_patients.plst'
geneFile = None #'/Users/jlu96/conte/jlu/REQUIREDFILES_OnlyLoss2/COSMICGenes_OnlyLoss.txt'
load_directory = '/Users/jlu96/conte/jlu/Analyses/CooccurImprovement/LoadMatrices'
minFreq = 0
num_permutations = 20
binary_perm_method = False
Q = 100
write_matrices = True
matrixdirectory = '/Users/jlu96/conte/jlu/Analyses/CooccurImprovement/LoadMatrices'
#'/Users/jlu96/conte/jlu/Analyses/CooccurImprovement/SARC_broad-som-jl-' + ('matrix' if binary_perm_method else 'network')
outmutexfile = matrixdirectory + '/mutex' + str(num_permutations) + str(time.time()) + '.tsv'
outcooccurfile = matrixdirectory + '/cooccur' + str(num_permutations) + str(time.time()) + '.tsv'
outseedsfile = matrixdirectory + '/seeds' + str(time.time()) + '.tsv'
if not os.path.exists(os.path.dirname(matrixdirectory)):
os.makedirs(os.path.dirname(matrixdirectory))
numGenes, numCases, genes, patients, geneToCases, patientToGenes = mex.load_mutation_data(mutationmatrix, patientFile, geneFile, minFreq)
print "numGenes ", numGenes, " and numCases ", numCases
for patient in patients:
if not patientToGenes[patient]:
patientToGenes.pop(patient)
print patient, "popped"
# Generate Permutation Matrices
pm = PermutationMatrices(geneToCases, patientToGenes, num_permutations, Q=Q, matrixdirectory=matrixdirectory,
binary_perm_method=binary_perm_method, write_matrices=write_matrices, load_directory=load_directory,
geneFile=geneFile, patientFile=patientFile, minFreq=minFreq)
# Make list of pairs from highly mutated genes
test_genes = [gene for gene in genes if len(geneToCases[gene]) > 5]
# for test_gene in test_genes:
# print test_gene
genepairs = met.getgenepairs(geneToCases, test_genes)
print "Number of pairs to test ", len(genepairs)
# CALCULATE MUTEX
# Create a list of ConditionFunctions that you must later initialize...
ConditionFunctions = range(len(genepairs))
mutex_set_condition_function_list = []
# Generate set_condition_function_list
for i in range(len(genepairs)):
genepair = genepairs[i]
condition_dict = {}
condition_dict['Genes'] = tuple(genepair)
condition_dict['Overlap'] = len(set.intersection(*[geneToCases[gene] for gene in condition_dict['Genes']]))
condition_dict['Mutex'] = True
ConditionFunctions[i] = Condition([condition_dict])
if [condition_dict] != ConditionFunctions[i].conditions:
print condition_dict, ConditionFunctions[i].conditions
mutex_set_condition_function_list.append((genepair, ConditionFunctions[i]))
print "Finished mutex condition function list"
t= time.time()
# Calculate pvalues for mutual exclusivity
pair_to_mutex = {}
pair_to_mutex_network_pvalue = pm.set_to_pvalue(mutex_set_condition_function_list)
print "mutex pair network pvalues finished in ", time.time() - t
for genepair in genepairs:
pair_to_mutex[genepair] = mex.analyze_mutex_set_new(numCases, geneToCases, patientToGenes, genepair)
pair_to_mutex[genepair]['NetworkProbability'] = pair_to_mutex_network_pvalue[genepair]
# Write to output
with open(outmutexfile, 'w') as csvfile:
fieldnames = pair_to_mutex[genepairs[0]].keys()
writer = csv.DictWriter(csvfile, delimiter='\t', fieldnames=fieldnames)
writer.writeheader()
for genepair in pair_to_mutex:
writer.writerow(pair_to_mutex[genepair])
# CALCULATE COOCCUR
cooccur_set_condition_function_list = []
# Generate set_condition_function_list
for genepair in genepairs:
ConditionFunction = Condition(None)
condition_dict = {}
condition_dict['Genes'] = tuple(genepair)
condition_dict['Overlap'] = len(set.intersection(*[geneToCases[gene] for gene in condition_dict['Genes']]))
condition_dict['Mutex'] = False
ConditionFunction.set_params([condition_dict])
cooccur_set_condition_function_list.append((genepair, ConditionFunction))
t= time.time()
# Calculate pvalues for mutual exclusivity
pair_to_cooccur = {}
pair_to_cooccur_network_pvalue = pm.set_to_pvalue(cooccur_set_condition_function_list)
print "cooccur pair network pvalues finished in ", time.time() - t
for genepair in genepairs:
pair_to_cooccur[genepair] = mex.analyze_cooccur_set_new(numCases, geneToCases, patientToGenes, genepair)
pair_to_cooccur[genepair]['NetworkProbability'] = pair_to_cooccur_network_pvalue[genepair]
# Write to output
with open(outcooccurfile, 'w') as csvfile:
fieldnames = pair_to_cooccur[genepairs[0]].keys()
writer = csv.DictWriter(csvfile, delimiter='\t', fieldnames=fieldnames)
writer.writeheader()
for genepair in pair_to_cooccur:
writer.writerow(pair_to_cooccur[genepair])
# Write seeds to output
with open(outseedsfile, 'w') as csvfile:
writer = csv.writer(csvfile, delimiter='\t')
for seed in pm.seeds:
writer.writerow([seed])
def complete_cooccurpairs(self, genepairs, p=0.05, minCooccur=1, min_cooccurrence_ratio=0.0, parallel_compute_number=0,
compute_scores=True):
"""
:param genepairs:
:param cprob:
:param minCooccur:
:param min_cooccurrence_ratio:
:param parallel_compute_number:
:return: cpairsdict, cgenedict
"""
print "Generating list of", len(genepairs), " co-occurring hypotheses to test on permutation matrices..."
# Generate condition functions after analyzing each gene pair for Co-occurrence/min
cooccur_set_condition_function_list = []
# Generate list of condition functions to test the permutation matrix for
for genepair in genepairs:
ConditionFunction = Condition(None)
condition_dict = {}
condition_dict['Genes'] = tuple(genepair)
condition_dict['Overlap'] = len(set.intersection(*[self.geneToCases_orig[gene] for gene in condition_dict['Genes']]))
condition_dict['Mutex'] = False
ConditionFunction.set_params([condition_dict])
cooccur_set_condition_function_list.append((genepair, ConditionFunction))
print "Done. Now, calulating p-values of hypotheses..."
# Generate co-occurring pairs
if parallel_compute_number:
cooccur_pair_to_pvalue = pac.parallel_compute_new(self.set_to_pvalue, [cooccur_set_condition_function_list],
cooccur_set_condition_function_list, 0, pac.partition_inputs, {0: pac.combine_dictionaries},
number=parallel_compute_number,
procnumber=parallel_compute_number)
else:
cooccur_pair_to_pvalue = self.set_to_pvalue(cooccur_set_condition_function_list)
print "Done. Now, finding co-occurring pairs"
# Generate dictionary for each pair. Optionally analyze each cooccur set as well.
cpairsdict = {}
cgenedict = {}
for genepair in cooccur_pair_to_pvalue:
if cooccur_pair_to_pvalue[genepair] < p:
cstats = mex.analyze_cooccur_set_new(self.numCases, self.geneToCases_orig, self.patientToGenes_orig,
geneset=tuple(genepair), compute_scores=compute_scores)
if cstats['Overlap'] >= minCooccur and cstats['CooccurrenceRatio'] >= min_cooccurrence_ratio:
cstats['PermutationProbability'] = cooccur_pair_to_pvalue[genepair]
cpairsdict[genepair] = cstats
gene1, gene2 = tuple(genepair)
if gene1 not in cgenedict:
cgenedict[gene1] = set()
cgenedict[gene1].add(gene2)
else:
cgenedict[gene1].add(gene2)
if gene2 not in cgenedict:
cgenedict[gene2] = set()
cgenedict[gene2].add(gene1)
else:
cgenedict[gene2].add(gene1)
return cpairsdict, cgenedict