def computeOverlapPval(s1, s2, n):
a = len(s1 & s2)
b = len(s1 - s2)
c = len(s2 - s1)
d = n - (a + b + c)
probability_value = fisher.compute(a, b, c, d)
return fisher.significance(probability_value, a, b, c, d)
def computeTraitStatistics(genes, pfilter):
traitSet = set([key for key in gwasDB.__studyByTrait if len(gwasDB.getGenesForTrait(key)) > 0])
traitChi = {}
pbar = ProgressBar()
pbar.setMaximum(len(traitSet))
pbar.updateProgress(0)
i=0
for trait in traitSet:
if i % 5 == 0:
pbar.updateProgress(0)
i+=1
traitGenes = gwasDB.getGenesForTrait(trait,pfilter)
listA = traitGenes & genes
listC = traitGenes - genes
a = len(listA)
b = len(genes - traitGenes)
c = len(listC)
d = len(geneDB.__approved_symbols - (traitGenes | genes))
oddsratio = geneUtils.oddsRatio(a,b,c,d)
kappa = geneUtils.kappaStatistic(a,b,c,d)
fisher_exact = fisher.compute(a,b,c,d)
fisher_p = fisher.significance(fisher_exact, a,b,c,d)
traitChi[trait] = (a, oddsratio, kappa, len(traitGenes),
fisher_exact, fisher_p, traitGenes)
pbar.finalize()
return traitChi
def computeTraitDrugLists(RE_genes, drug_genes, pfilter_cutoff):
traitSet = set(gwasDB.__studyByTrait.keys())
for trait in traitSet:
traitGenes = gwasDB.getGenesForTrait(trait, pfilter_cutoff)
if len(traitGenes) == 0:
continue
RE_drugs = []
other_drugs = []
drug_counts_by_gene = {}
drugs_targeting_RE = set([])
allDrugs = set([])
for gene in traitGenes & RE_genes:
drug_counts_by_gene[gene] = 0
if gene in drugDB.__drugDict:
for drug in drugDB.__drugDict[gene]:
RE_drugs.append(drug)
drug_counts_by_gene[gene] += len(drugDB.__drugDict[gene])
drugs_targeting_RE |= drugDB.__drugDict[gene]
drugs_targeting_disease = set([])
for gene in ((traitGenes & drug_genes) - RE_genes):
drug_counts_by_gene[gene] = 0
if gene in drugDB.__drugDict:
for drug in drugDB.__drugDict[gene]:
other_drugs.append(drug)
drug_counts_by_gene[gene] += len(drugDB.__drugDict[gene])
drugs_targeting_disease |= drugDB.__drugDict[gene]
drugs_targeting_other_RE = set([])
for gene in RE_genes:
drug_counts_by_gene[gene] = 0
if gene in drugDB.__drugDict:
drugs_targeting_other_RE |= drugDB.__drugDict[gene]
drugs_targeting_other_RE -= drugs_targeting_RE
a = len(drugs_targeting_RE)
b = len(drugs_targeting_disease)
c = len(drugs_targeting_other_RE)
d = len(set(drugDB.__drugs.keys()) - (drugs_targeting_RE |
drugs_targeting_disease | drugs_targeting_other_RE))
# print a, b, c, d
if (a + b) == 0 or (a + c) == 0:
odds, kappa = 0, 0
fisher_exact = 1.0
fisher_p = 1.0
else:
odds = geneUtils.oddsRatio(a,b,c,d)
kappa = geneUtils.kappaStatistic(a,b,c,d)
fisher_exact = fisher.compute(a,b,c,d)
fisher_p = fisher.significance(fisher_exact, a,b,c,d)
__traitMetaAnalysis[trait]['RE_drugs'] = RE_drugs
__traitMetaAnalysis[trait]['other_drugs'] = other_drugs
__traitMetaAnalysis[trait]['drug_counts'] = drug_counts_by_gene
__traitMetaAnalysis[trait]['drugchi'] = (a,b,c,d,odds,kappa,fisher_exact,fisher_p)
def computeTraitGeneLists(RE_genes, drug_genes, pfilter_cutoff):
traitSet = set(gwasDB.__studyByTrait.keys())
pbar = ProgressBar()
pbar.setMaximum(len(traitSet))
pbar.updateProgress(0)
i = 0
for trait in traitSet:
if i % 5 == 0:
pbar.updateProgress(i)
i+=1
traitGenes = gwasDB.getGenesForTrait(trait, pfilter_cutoff)
if len(traitGenes) == 0:
continue
__traitMetaAnalysis[trait] = {}
RE = []
for gene in traitGenes & RE_genes:
count = len(gwasDB.getTraitsForGene(gene))
RE.append((gene, count))
__traitMetaAnalysis[trait]['RE'] = RE
drug = []
for gene in traitGenes & drug_genes:
count = len(gwasDB.getTraitsForGene(gene))
drug.append((gene, count))
__traitMetaAnalysis[trait]['drugbank'] = drug
other = []
for gene in traitGenes - RE_genes - drug_genes:
count = len(gwasDB.getTraitsForGene(gene))
other.append((gene,count))
__traitMetaAnalysis[trait]['other'] = other
a = len(traitGenes & RE_genes)
b = len(RE_genes - traitGenes)
c = len(traitGenes - RE_genes)
d = len(geneDB.__approved_symbols - (traitGenes | RE_genes))
oddsratio = geneUtils.oddsRatio(a,b,c,d)
kappa = geneUtils.kappaStatistic(a,b,c,d)
fisher_exact = fisher.compute(a,b,c,d)
fisher_p = fisher.significance(fisher_exact, a,b,c,d)
__traitMetaAnalysis[trait]['RE_chi'] = (a, b, c, d,
oddsratio, kappa, fisher_exact, fisher_p)
a = len(traitGenes & drug_genes)
b = len(drug_genes - traitGenes)
c = len(traitGenes - drug_genes)
d = len(geneDB.__approved_symbols - (traitGenes | drug_genes))
oddsratio = geneUtils.oddsRatio(a,b,c,d)
kappa = geneUtils.kappaStatistic(a,b,c,d)
fisher_exact = fisher.compute(a,b,c,d)
fisher_p = fisher.significance(fisher_exact, a,b,c,d)
__traitMetaAnalysis[trait]['drugbank_chi'] = (a, b, c, d,
oddsratio, kappa, fisher_exact, fisher_p)
__traitMetaAnalysis[trait]['geneset_size'] = len(traitGenes)
pbar.finalize()
cgWithoutGWAS = studyGenes - gwasDB.__geneSet gwasWithoutCG = gwasDB.__geneSet - studyGenes print "initializing fisher algorithm..." fisher.init(len(geneDB.__approved_symbols)) # chi-square contingency table for gwas and RE a1 = len(commonGenes) b1 = len(gwasWithoutCG) c1 = len(cgWithoutGWAS) d1 = len( geneDB.__approved_symbols - (studyGenes | gwasDB.__geneSet) ) oddsratio1 = geneUtils.oddsRatio(a1,b1,c1,d1) kappa1 = geneUtils.kappaStatistic(a1,b1,c1,d1) fisher_exact1 = fisher.compute(a1,b1,c1,d1) fisherp1 = fisher.significance(fisher_exact1, a1,b1,c1,d1) commonDrugTargets = drugDB.__geneSet & studyGenes a2 = len( commonDrugTargets ) b2 = len( drugDB.__geneSet - studyGenes ) c2 = len( studyGenes - drugDB.__geneSet ) d2 = len( geneDB.__approved_symbols - ( studyGenes | drugDB.__geneSet ) ) oddsratio2 = geneUtils.oddsRatio(a2,b2,c2,d2) kappa2 = geneUtils.kappaStatistic(a2,b2,c2,d2) fisher_exact2 = fisher.compute(a2,b2,c2,d2) fisherp2 = fisher.significance(fisher_exact2, a2,b2,c2,d2) gwas_drugbank_overlap = gwasDB.__geneSet & drugDB.__geneSet