def simultaneousPermutationWithMultipleSignificanceLevels(
significance_levels, n, categories, testset_size, iterations):
print "Running permutation test on %d categories with test set size of %d drawn from %d items for %d iterations" % (
len(categories), testset_size, n, iterations)
pbar = ProgressBar()
pbar.setMaximum(iterations)
pbar.updateProgress(0)
total_sig_hypotheses = [0] * len(significance_levels)
num_sig_levels = len(significance_levels)
sampleset = xrange(0, n)
for i in xrange(0, iterations):
if i % 1 == 0:
pbar.updateProgress(i)
test_set = set(random.sample(sampleset, testset_size))
for j, s_c in enumerate(categories):
pval = computeOverlapPval(s_c, test_set, n)
for k in xrange(0, num_sig_levels):
if pval <= significance_levels[k]:
total_sig_hypotheses[k] += 1
pbar.finalize()
return [
float(total_sig) / float(iterations)
for total_sig in total_sig_hypotheses
]
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 createGeneListingsHTML(geneListDir):
pbar = ProgressBar()
pbar.setMaximum(len(gwasDB.__geneSet))
pbar.updateProgress(0)
i=0
for gene in gwasDB.__geneSet:
if i % 10 == 0:
pbar.updateProgress(i)
i+=1
genePage = htmltools.createPage("Gene Summary: " +
geneDB.__original_names[gene], css_file='../genereport.css',
scripts = {'../sorttable.js':'javascript'})
# Create the disease trait tables
traits = []
for trait in gwasDB.__traitDict[gene]:
traits.append(trait)
traits = sorted(traits, key=lambda trait: -__traitMetaAnalysis[trait]['RE_chi'][0])
traitTable = []
for trait in traits:
cnt = len(__traitMetaAnalysis[trait]['RE'])
oddsratio = __traitMetaAnalysis[trait]['RE_chi'][4]
kappa = __traitMetaAnalysis[trait]['RE_chi'][5]
fisher_exact = __traitMetaAnalysis[trait]['RE_chi'][6]
fisherp = __traitMetaAnalysis[trait]['RE_chi'][7]
numgenes = __traitMetaAnalysis[trait]['geneset_size']
translate = trait.replace(" ","_").replace("/", " or ").replace("\\", " or ")
if len(trait) > 38:
trait = trait[:35] + "..."
traitTable.append(["<a href=\"../traitlists/%s.html\">%s</a>" %
(translate,trait), cnt, numgenes, "%.7f" % (fisher_exact),
"%.7f" % (fisherp), "%.1f" % (oddsratio), "%.4f" % (kappa), ])
genePage.div("Gene %s, total traits: %d" % (geneDB.__original_names[gene],len(traitTable)), class_="header")
htmltools.createTable(genePage, traitTable, ["Disease/Trait", "#RE Genes",
"#Trait Genes", "fisher exact", "P-value",
"oddsratio", "kappa"], "traitlisthead", None,
["traitcol","recol",
"genecol","fishercol","pcol",
"oddscol","kappacol"], "sortable", None)
# Create drug bank links
if gene not in drugDB.__drugDict:
genePage.div("No drugs target gene %s" % (geneDB.__original_names[gene]), class_="header")
else:
drugbank_size = len(drugDB.__drugDict[gene])
genePage.div("%d drugs targeting gene %s" % (drugbank_size, geneDB.__original_names[gene]), class_="header")
genePage.div.open(class_="druglist")
genePage.ul.open()
for drug in drugDB.__drugDict[gene]:
link = "http://www.drugbank.ca/drugs/%s" % (drug)
if drug not in drugDB.__drugs:
genePage.li(oneliner.a(drug, href=link))
else:
genePage.li(oneliner.a(drugDB.__drugs[drug]['name'], href=link))
genePage.ul.close()
genePage.div.close()
htmltools.savePage(genePage, os.sep.join([geneListDir, gene + ".html"]))
pbar.finalize()
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()
