def rankSomaticMutations_byTreatmentContingency(inputRNAFilename, outputFilename, treatmentsCSVFilename, additionalHeaders = ("Entrez_Gene_Id","Chrom","Start_Position","End_Position","Strand","Variant_Type","Reference_Allele","Tumor_Seq_Allele1","Tumor_Seq_Allele2"), genesHeaderColumnName = "Hugo_Symbol", delimiter = "\t"):
'''
Contingency table approach for measuring significant difference in somatic mutations between different treatment groups.
The code builds a contingency table for each treatment pair using counts of mutated and non-mutated genes for each treatment.
Use the additionalHeaders parameter to add additional headers that may occur in mutation signiture files.
'''
from Utility.Utility import Utility
import numpy as np
import scipy.stats as scipystats
USE_NUMPY_WRAPPING = True
assert treatmentsCSVFilename
assert inputRNAFilename
assert outputFilename
treatmentsDict = Analysis.buildTreatmentsLists(treatmentsCSVFilename, delimiter)
if treatmentsDict not in ({},None):
outputFile = open(outputFilename, "w")
headerLine = genesHeaderColumnName + delimiter
if additionalHeaders not in (None, ()):
for addHeader in additionalHeaders:
headerLine += addHeader + delimiter
for treat1 in treatmentsDict.keys():
for treat2 in treatmentsDict.keys():
if treat1 != treat2:
headerLine += "%s_%s_pValue" % (treat1, treat2) + delimiter
outputFile.write(headerLine.strip(delimiter) + "\n")
with open(inputRNAFilename, "r") as inputRNAFile:
columns, indexToName = Utility.getColumns(inputRNAFile) # @UnusedVariable
geneList = columns[genesHeaderColumnName]
totalMutations = len(geneList)
counter = 0
for i, gene in enumerate(geneList):
counter += 1
if counter % 1000 == 0:
print "Progress completed: %s percent. Working on mutation at gene: %s" % (str((float(counter)/totalMutations)*100), gene)
geneOutputLine = gene + delimiter
if additionalHeaders not in (None, ()):
for addHeader in additionalHeaders:
geneOutputLine += columns[addHeader][i] + delimiter
samplesDict = {}
for treatment in treatmentsDict.keys():
samplesDict[treatment] = []
samplesPerTreatment = treatmentsDict[treatment]
for sampleID in samplesPerTreatment:
try:
samplesDict[treatment].append(columns[sampleID][i])
except:
#print "Sample not found sampleID: %s" % (sampleID)
pass
if USE_NUMPY_WRAPPING:
samplesDict[treatment] = np.array(samplesDict[treatment]).astype(np.float)
for treat1 in treatmentsDict.keys():
for treat2 in treatmentsDict.keys():
if treat1 != treat2:
p = float(1.0)
try:
p = Analysis.calculateContingencyP_expectedWeighted(samplesDict[treat1], samplesDict[treat2])
geneOutputLine += str(p) + delimiter
except:
geneOutputLine += str(p) + delimiter
outputFile.write(geneOutputLine.strip(delimiter) + "\n")
outputFile.flush()
outputFile.close()
return outputFilename
def rankRNAExpressionTreatments_byMannWhitney(inputRNAFilename, outputFilename, treatmentsCSVFilename, genesHeaderColumnName = "gene_id|gene_id_code", delimiter = "\t"):
'''
Method uses Mann-Whitney test to determine variations in the distributions of RNA expression data between treatments.
'''
from Utility.Utility import Utility
import numpy as np
import scipy.stats as scistats
TEST_TYPE = "mannwhitneyu"
USE_NUMPY_WRAPPING = True
assert treatmentsCSVFilename
assert inputRNAFilename
assert outputFilename
treatmentsDict = Analysis.buildTreatmentsLists(treatmentsCSVFilename, delimiter)
if treatmentsDict not in ({},None):
outputFile = open(outputFilename, "w")
headerLine = "gene_id" + delimiter
for treat1 in treatmentsDict.keys():
for treat2 in treatmentsDict.keys():
if treat1 != treat2:
headerLine += "%s_%s_U_%s" % (treat1, treat2, TEST_TYPE) + delimiter
headerLine += "%s_%s_p_%s" % (treat1, treat2, TEST_TYPE) + delimiter
outputFile.write(headerLine.strip(delimiter) + "\n")
with open(inputRNAFilename, "r") as inputRNAFile:
columns, indexToName = Utility.getColumns(inputRNAFile) # @UnusedVariable
geneList = columns[genesHeaderColumnName]
if geneList[0].find("|") != -1:
geneList = [x.split("|")[0] for x in geneList]
for i, gene in enumerate(geneList):
geneOutputLine = gene + delimiter
samplesDict = {}
for treatment in treatmentsDict.keys():
samplesDict[treatment] = []
samplesPerTreatment = treatmentsDict[treatment]
for sampleID in samplesPerTreatment:
try:
samplesDict[treatment].append(columns[sampleID][i])
except:
#print "Sample not found sampleID: %s" % (sampleID)
pass
if USE_NUMPY_WRAPPING:
samplesDict[treatment] = np.array(samplesDict[treatment]).astype(np.float)
for treat1 in treatmentsDict.keys():
for treat2 in treatmentsDict.keys():
if treat1 != treat2:
pValue = float(1.0)
U = float(0.0)
try:
U, pValue = scistats.mannwhitneyu(samplesDict[treat1], samplesDict[treat2]) # @UnusedVariable
geneOutputLine += str(U) + delimiter
geneOutputLine += str(pValue) + delimiter
except:
geneOutputLine += str(U) + delimiter
geneOutputLine += str(pValue) + delimiter
outputFile.write(geneOutputLine.strip(delimiter) + "\n")
outputFile.flush()
outputFile.close()
return outputFilename
def WriteVRTransaction(set, writeHeader):
transactions, statement, nonUpdatedStatementExporter, simpleStatementExporter = ReadTransaction(set, writeHeader);
vrFormat = ValueResearchStatement(statement, writeHeader);
Utility.WriteStatement(vrFormat, set.outputVRFormat);