def _plotKinshipDiffs_():
filterProb = 0.2
p_i = 1
res_dir = "/Users/bjarni/tmp/"
runId = "full_"
snpsDataFile = "/Network/Data/250k/dataFreeze_080608/250K_f10_080608.csv"
snpsds = dataParsers.parseCSVData(snpsDataFile, format=1, deliminator=",") # ,debug=True)
phenotypeFile = "/Network/Data/250k/dataFreeze_080608/phenotypes_all_raw_111008.tsv"
print "Loading phenotype data"
phed = phenotypeData.readPhenotypeFile(phenotypeFile, delimiter="\t")
snpsdata.coordinateSnpsAndPhenotypeData(phed, p_i, snpsds)
for snpsd in snpsds:
snpsd.filterMinMAF(0.1)
snpsd.filterMonoMorphicSnps()
totalSNPs = []
for i in range(len(snpsds)):
snpsds[i] = snpsds[i].getSnpsData()
totalSNPs += snpsds[i].snps
# For memory, remove random SNPs
snps = []
for snp in totalSNPs:
if random.random() < filterProb:
snps.append(snp)
totalSNPs = snps
print "Calculating the global kinship..."
globalKinship = calcKinship(totalSNPs)
print "done."
normalizedGlobalKinship = globalKinship / mean(globalKinship)
gc.collect() # Calling garbage collector, in an attempt to clean up memory..
for i in range(4, 5): # len(snpsds)):
chr = i + 1
snpsd = snpsds[i]
# pylab.subplot(5,1,chr)
# pylab.figure(figsize=(18,4))
# (kinshipDiffs,binPos,local300Kinships) = getKinshipDiffs(snpsd,normalizedGlobalKinship,windowSize=300000)
# pylab.plot(binPos,kinshipDiffs,"r",label='ws$=300000$')
# (kinshipDiffs,binPos,local500Kinships) = getKinshipDiffs(snpsd,normalizedGlobalKinship,windowSize=500000)
# pylab.plot(binPos,kinshipDiffs,"b",label='ws$=500000$')
# pylab.legend(numpoints=2,handlelen=0.005)
# pylab.title("Kinship diff. chr. "+str(chr))
# pylab.savefig(res_dir+runId+"kinshipDiffs_500_300kb_chr"+str(chr)+".pdf",format="pdf")
# pylab.clf()
pylab.figure(figsize=(18, 4))
(emmaDiffs, binPos) = getEmmaDiffs(snpsd, phed, p_i, globalKinship, windowSize=300000)
pylab.plot(binPos, emmaDiffs, "r", label="ws$=300000$")
pylab.title("Emma avg. p-value diff. 500kb on chr. " + str(chr))
(emmaDiffs, binPos) = getEmmaDiffs(snpsd, phed, p_i, globalKinship, windowSize=500000)
pylab.plot(binPos, emmaDiffs, "b", label="ws$=500000$")
pylab.title("Emma avg. p-value diff. on chr. " + str(chr))
pylab.legend(numpoints=2, handlelen=0.005)
pylab.savefig(res_dir + runId + "EmmaPvalDiffs_500_300kb_chr" + str(chr) + ".pdf", format="pdf")
pylab.clf()
gc.collect() # Calling garbage collector, in an attempt to clean up memory..
def _plotKW_():
"""
Analyze how population structure affects KW.
"""
filterProb = 0.1
p_i = 1
res_dir = "/Users/bjarni/tmp/"
runId = "_full_quick_"
snpsDataFile = "/Network/Data/250k/dataFreeze_080608/250K_f10_080608.csv"
snpsds = dataParsers.parseCSVData(snpsDataFile, format=1, deliminator=",") # ,debug=True)
phenotypeFile = "/Network/Data/250k/dataFreeze_080608/phenotypes_all_raw_111008.tsv"
print "Loading phenotype data"
phed = phenotypeData.readPhenotypeFile(phenotypeFile, delimiter="\t")
snpsdata.coordinateSnpsAndPhenotypeData(phed, p_i, snpsds)
totalSNPs = []
for i in range(len(snpsds)):
snpsds[i] = snpsds[i].getSnpsData()
totalSNPs += snpsds[i].snps
# For memory, remove random SNPs
snps = []
for snp in totalSNPs:
if random.random() < filterProb:
snps.append(snp)
totalSNPs = snps
# globalKinship = calcKinship(totalSNPs)
gc.collect() # Calling garbage collector, in an attempt to clean up memory..
# chr = 1
# for snpsd in snpsds:
snpsd = snpsds[3]
k = calcKinship(snpsd.snps[200:1400])
res = runEmma(phed, p_i, k, snpsd.snps[200:1400]) # runEmma(phed,p_i,k,snps):
pvals = res["ps"]
log_pvals = []
for pval in pvals:
# print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400], log_pvals, "c.", label="Emma (local)")
k = calcKinship(totalSNPs)
res = runEmma(phed, p_i, k, snpsd.snps[200:1400]) # runEmma(phed,p_i,k,snps):
pvals = res["ps"]
log_pvals = []
for pval in pvals:
# print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400], log_pvals, "g.", label="Emma (global)")
phenVals = phed.getPhenVals(p_i)
pvals = _run_kw_(snpsd.snps[200:1400], phenVals)
log_pvals = []
for pval in pvals:
# print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400], log_pvals, "r.", label="KW (full data)")
(pvals, new_positions, acc_groups) = get_KW_pvals(
snpsd.snps[200:1400], snpsd.positions[200:1400], phed, p_i, kinshipThreshold=0.95, method="KW"
)
ecot_map = phenotypeData._getEcotypeIdToStockParentDict_()
for i in range(0, len(acc_groups)):
acc_list = []
for a_i in acc_groups[i]:
e_i = snpsd.accessions[a_i]
# print e_i
acc_list.append(ecot_map[int(e_i)][0])
print "group", i, ":", acc_list
log_pvals = []
for pval in pvals:
# print pval
log_pvals.append(-math.log10(pval))
pylab.plot(new_positions, log_pvals, "b.", label="KW (merged data)")
pylab.legend(numpoints=2, handlelen=0.005)
pylab.show()
def _plotKW_():
"""
Analyze how population structure affects KW.
"""
filterProb = 0.1
p_i = 1
res_dir = "/Users/bjarni/tmp/"
runId = "_full_quick_"
snpsDataFile = "/Network/Data/250k/dataFreeze_080608/250K_f10_080608.csv"
snpsds = dataParsers.parseCSVData(snpsDataFile, format=1,
deliminator=",") #,debug=True)
phenotypeFile = "/Network/Data/250k/dataFreeze_080608/phenotypes_all_raw_111008.tsv"
print "Loading phenotype data"
phed = phenotypeData.readPhenotypeFile(phenotypeFile, delimiter='\t')
snpsdata.coordinateSnpsAndPhenotypeData(phed, p_i, snpsds)
totalSNPs = []
for i in range(len(snpsds)):
snpsds[i] = snpsds[i].getSnpsData()
totalSNPs += snpsds[i].snps
#For memory, remove random SNPs
snps = []
for snp in totalSNPs:
if random.random() < filterProb:
snps.append(snp)
totalSNPs = snps
#globalKinship = calcKinship(totalSNPs)
gc.collect(
) #Calling garbage collector, in an attempt to clean up memory..
#chr = 1
#for snpsd in snpsds:
snpsd = snpsds[3]
k = calcKinship(snpsd.snps[200:1400])
res = runEmma(phed, p_i, k,
snpsd.snps[200:1400]) #runEmma(phed,p_i,k,snps):
pvals = res["ps"]
log_pvals = []
for pval in pvals:
#print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400],
log_pvals,
"c.",
label="Emma (local)")
k = calcKinship(totalSNPs)
res = runEmma(phed, p_i, k,
snpsd.snps[200:1400]) #runEmma(phed,p_i,k,snps):
pvals = res["ps"]
log_pvals = []
for pval in pvals:
#print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400],
log_pvals,
"g.",
label="Emma (global)")
phenVals = phed.getPhenVals(p_i)
pvals = _run_kw_(snpsd.snps[200:1400], phenVals)
log_pvals = []
for pval in pvals:
#print pval
log_pvals.append(-math.log10(pval))
pylab.plot(snpsd.positions[200:1400],
log_pvals,
"r.",
label="KW (full data)")
(pvals, new_positions,
acc_groups) = get_KW_pvals(snpsd.snps[200:1400],
snpsd.positions[200:1400],
phed,
p_i,
kinshipThreshold=0.95,
method="KW")
ecot_map = phenotypeData._getEcotypeIdToStockParentDict_()
for i in range(0, len(acc_groups)):
acc_list = []
for a_i in acc_groups[i]:
e_i = snpsd.accessions[a_i]
#print e_i
acc_list.append(ecot_map[int(e_i)][0])
print "group", i, ":", acc_list
log_pvals = []
for pval in pvals:
#print pval
log_pvals.append(-math.log10(pval))
pylab.plot(new_positions, log_pvals, "b.", label="KW (merged data)")
pylab.legend(numpoints=2, handlelen=0.005)
pylab.show()
def _plotKinshipDiffs_():
filterProb = 0.2
p_i = 1
res_dir = "/Users/bjarni/tmp/"
runId = "full_"
snpsDataFile = "/Network/Data/250k/dataFreeze_080608/250K_f10_080608.csv"
snpsds = dataParsers.parseCSVData(snpsDataFile, format=1,
deliminator=",") #,debug=True)
phenotypeFile = "/Network/Data/250k/dataFreeze_080608/phenotypes_all_raw_111008.tsv"
print "Loading phenotype data"
phed = phenotypeData.readPhenotypeFile(phenotypeFile, delimiter='\t')
snpsdata.coordinateSnpsAndPhenotypeData(phed, p_i, snpsds)
for snpsd in snpsds:
snpsd.filterMinMAF(0.1)
snpsd.filterMonoMorphicSnps()
totalSNPs = []
for i in range(len(snpsds)):
snpsds[i] = snpsds[i].getSnpsData()
totalSNPs += snpsds[i].snps
#For memory, remove random SNPs
snps = []
for snp in totalSNPs:
if random.random() < filterProb:
snps.append(snp)
totalSNPs = snps
print "Calculating the global kinship..."
globalKinship = calcKinship(totalSNPs)
print "done."
normalizedGlobalKinship = globalKinship / mean(globalKinship)
gc.collect(
) #Calling garbage collector, in an attempt to clean up memory..
for i in range(4, 5): #len(snpsds)):
chr = i + 1
snpsd = snpsds[i]
#pylab.subplot(5,1,chr)
# pylab.figure(figsize=(18,4))
# (kinshipDiffs,binPos,local300Kinships) = getKinshipDiffs(snpsd,normalizedGlobalKinship,windowSize=300000)
# pylab.plot(binPos,kinshipDiffs,"r",label='ws$=300000$')
# (kinshipDiffs,binPos,local500Kinships) = getKinshipDiffs(snpsd,normalizedGlobalKinship,windowSize=500000)
# pylab.plot(binPos,kinshipDiffs,"b",label='ws$=500000$')
# pylab.legend(numpoints=2,handlelen=0.005)
# pylab.title("Kinship diff. chr. "+str(chr))
# pylab.savefig(res_dir+runId+"kinshipDiffs_500_300kb_chr"+str(chr)+".pdf",format="pdf")
# pylab.clf()
pylab.figure(figsize=(18, 4))
(emmaDiffs, binPos) = getEmmaDiffs(snpsd,
phed,
p_i,
globalKinship,
windowSize=300000)
pylab.plot(binPos, emmaDiffs, "r", label='ws$=300000$')
pylab.title("Emma avg. p-value diff. 500kb on chr. " + str(chr))
(emmaDiffs, binPos) = getEmmaDiffs(snpsd,
phed,
p_i,
globalKinship,
windowSize=500000)
pylab.plot(binPos, emmaDiffs, "b", label='ws$=500000$')
pylab.title("Emma avg. p-value diff. on chr. " + str(chr))
pylab.legend(numpoints=2, handlelen=0.005)
pylab.savefig(res_dir + runId + "EmmaPvalDiffs_500_300kb_chr" +
str(chr) + ".pdf",
format="pdf")
pylab.clf()
gc.collect(
) #Calling garbage collector, in an attempt to clean up memory..