def execute_gw_regions(args):
''' apply significance cutoff to genome-wide data to identify regions '''
basedir = args.writedir
pop = args.emppop
thresshold = args.thresshold
cutoff = args.cutoff
windowlen = args.regionlen
suffix = args.suffix
chroms = range(1,23)
signif_windows = []
####################
## LOOP OVER CHRS ##
####################
for chrom in chroms:
chrom_signif = []
normedempfilename = get_emp_cms_file(pop, chrom, normed=True, suffix=suffix, basedir=basedir)
if not os.path.isfile(normedempfilename):
print("missing: " + normedempfilename)
else:
physpos, genpos, seldaf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(normedempfilename)
for iPos in range(len(physpos)):
##################
## CHECK REGION ##
##################
window_scores = get_window(iPos, physpos, cms_normed, windowlen)
percentage = check_outliers(window_scores, cutoff)
if percentage > thresshold:
chrom_signif.append(physpos[iPos])
signif_windows.append(chrom_signif)
##############################
## MERGE CONTIGUOUS WINDOWS ##
##############################
final_starts = []
final_ends = []
print('merging regions')
for chrom_signif in signif_windows:
starts, ends = merge_windows(chrom_signif, windowlen)
final_starts.append(starts)
final_ends.append(ends)
###################
## WRITE TO FILE ##
###################
if args.saveLog is not None:
writefilename = args.saveLog
writefile = open(writefilename, 'w')
for ichrom in range(len(final_starts)):
chromnum = ichrom + 1
starts = final_starts[ichrom]
ends = final_ends[ichrom]
for iregion in range(len(starts)-1):
writeline = "chr" + str(chromnum) + "\t" + str(starts[iregion]) + "\t" + str(ends[iregion]) + '\n'
writefile.write(writeline)
writefile.close()
print('wrote to ' + writefilename)
return
def execute_tpr(args):
''' estimate true positive rate for region detection '''
model = args.model
regionlen = args.regionlen
thresshold = args.thresshold
cutoff = args.cutoff
numReps = args.nrep
pop = args.simpop
suffix = args.suffix
writedir = args.writedir
takeScore = args.score
all_scores = []
all_percentages = []
#if args.saveLog is not None:
# writefilename = args.saveLog
# if os.path.isfile(writefilename):
# print(writefilename + " already exists; aborting.")
# sys.exit(0)
#per seldaf
dafbins = [['0.10', '0.20', '0.30', '0.40', '0.50', '0.60', '0.70', '0.80', '0.90'], ['0.10', '0.20', '0.30'], ['0.40', '0.50', '0.60'], ['0.70', '0.80', '0.90'], ['0.90']]
daflabels = ['all', 'lo', 'mid', 'hi','highest']
for ibin in [3]:#[1, 2, 3, 4]:#range(1):
thesebins, thislabel = dafbins[ibin], daflabels[ibin]
allrepfilenames = []
for selbin in thesebins:
for irep in range(1, numReps + 1):
repfilename = get_sel_repfile_name(model, irep, pop, selbin, normed=True, suffix=suffix, basedir=writedir)
if (irep==1):
print(repfilename)
if os.path.isfile(repfilename):
allrepfilenames.append(repfilename)
print('loaded ' + str(len(allrepfilenames)) + " replicates...")
#numToTake = min(500, len(allrepfilenames))
#chosen = np.random.choice(allrepfilenames, numToTake, replace=False) #take random sample
chosen = allrepfilenames #this was just to expedite, no?
for repfilename in chosen:
physpos, genpos, seldaf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(repfilename)
#physpos, genpos, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(repfilename)
these_scores = eval(takeScore)
if len(these_scores) > 0:
all_scores.append(these_scores)
rep_percentages = check_rep_windows(physpos, these_scores, regionlen, cutoff = cutoff)
all_percentages.append(rep_percentages)
print('loaded ' + str(len(all_scores)) + " replicates populations for model " + model + "...")
tpr = calc_pr(all_percentages, thresshold)
print('true positive rate: ' + str(tpr) + "\n")
if args.saveLog is not None:
writefilename = args.saveLog +"_" + thislabel
writefile = open(writefilename, 'w')
writefile.write(str(tpr)+'\n')
writefile.write(model + "\t" + str(regionlen) + "\t" + str(thresshold) + '\t' + str(cutoff) + '\n')
writefile.close()
print('wrote to : ' + str(writefilename))
return
def execute_fpr(args):
''' estimate false positive rate for region identification '''
model = args.model
regionlen = args.regionlen
thresshold = args.thresshold
cutoff = args.cutoff
numReps = args.nrep
pop = args.simpop
suffix = args.suffix
writedir = args.writedir
takeScore = args.score
all_scores = []
all_percentages = []
if True:
for irep in range(1, numReps + 1):
repfilename = get_neut_repfile_name(model, irep, pop, normed=True, suffix=suffix, basedir=writedir)
if (irep==1):
print(repfilename)
physpos, genpos, seldaf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(repfilename)
#physpos, genpos, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(repfilename)
these_scores = eval(takeScore)
if len(these_scores) > 0:
all_scores.append(these_scores)
rep_percentages = check_rep_windows(physpos, these_scores, regionlen, cutoff = cutoff)
all_percentages.append(rep_percentages)
#FOR DEBUG
#print(str(rep_percentages) + "\t" + repfilename)
if len(rep_percentages) > 0:
if max(rep_percentages) > thresshold:
print("false positive: " + repfilename)
print('loaded ' + str(len(all_scores)) + " replicates populations for model " + model + "...")
fpr = calc_pr(all_percentages, thresshold)
print('false positive rate: ' + str(fpr) + "\n")
if args.saveLog is not None:
writefilename = args.saveLog
writefile = open(writefilename, 'w')
writefile.write(str(fpr)+'\n')
writefile.write(model + "\t" + str(regionlen) + "\t" + str(thresshold) + '\t' + str(cutoff) + '\n')
writefile.close()
print('wrote to : ' + str(writefilename))
return
def execute_cdf(args):
""" visualize power to localize variants: estimate p(causal variant captured | signif thresshold includes x top SNPs) from simulates. plot as cumulative density function"""
reps = args.nrep
savefilename = args.savefilename
writedir = args.writedir
scenars = ['0.70', '0.80', '0.90']#'0.10', '0.20', '0.30', '0.40', '0.50', '0.60', '0.70', '0.80', '0.90']
model = args.model
causalPos = args.selPos
suffix = args.suffix
#causal_ranks_all = []
causal_ranks_1, causal_ranks_2, causal_ranks_3, causal_ranks_4 = [], [], [], []
for pop in [1, 2, 3, 4]:
for scenar in scenars:
for irep in range(1, reps+1):
cmsfilename = get_sel_repfile_name(model, irep, pop, scenar, normed = False, basedir=writedir, suffix=suffix)
if os.path.isfile(cmsfilename):
physpos, genpos, seldaf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(cmsfilename)
if causalPos in physpos:
causal_index = physpos.index(causalPos)
causal_unnormed = cms_unnormed[causal_index]
causal_rank = get_causal_rank(cms_unnormed, causal_unnormed)
#print(cmsfilename)
#print('causal rank: ' + str(causal_rank))
#causal_ranks.append(causal_rank)
this_array = eval('causal_ranks_' + str(pop))
if not np.isnan(causal_rank):
this_array.append(causal_rank)
else:
print("missing; " + cmsfilename)
print("for pop 1, loaded " + str(len(causal_ranks_1)) + " replicates.")
print("for pop 2, loaded " + str(len(causal_ranks_2)) + " replicates.")
print("for pop 3, loaded " + str(len(causal_ranks_3)) + " replicates.")
print("for pop 4, loaded " + str(len(causal_ranks_4)) + " replicates.")
cdf_fig, cdf_ax = plt.subplots()
if len(causal_ranks_1) > 0:
cdf_bins1, cdf1 = get_cdf_from_causal_ranks(causal_ranks_1)
cdf_ax.plot(cdf_bins1[1:], cdf1, color="yellow")
if len(causal_ranks_2) > 0:
cdf_bins2, cdf2 = get_cdf_from_causal_ranks(causal_ranks_2)
cdf_ax.plot(cdf_bins2[1:], cdf2, color="blue")
if len(causal_ranks_3) > 0:
cdf_bins3, cdf3 = get_cdf_from_causal_ranks(causal_ranks_3)
cdf_ax.plot(cdf_bins3[1:], cdf3, color="green")
if len(causal_ranks_4) > 0:
cdf_bins4, cdf4 = get_cdf_from_causal_ranks(causal_ranks_4)
cdf_ax.plot(cdf_bins4[1:], cdf4, color="purple")
cdf_ax.set_xlim([0, 50])
plt.title(model) #+ ", " + str(len(causal_ranks)) + " selection replicates")
plt.ylabel('probability that the causal variant is captured')
plt.xlabel('significance thresshold (i.e., examining the top x variants)')
plt.savefig(savefilename)
plt.close()
print('plotted to ' + savefilename)
return
def execute_extended_manhattan(args):
""" generate a genome-wide plot of CMS scores with option to hilight outlier regions """
plotscore = args.plotscore
selpop = args.emppop
basedir = args.writedir
suffix = args.suffix
savename = args.savefilename
dpi = args.dpi
numChr = 22
titlestring = args.titlestring
modelpops = {'YRI':1, 'GWD':1, 'LWK':1, 'MSL':1, 'ESN':1,
'CEU':2, 'FIN':2, 'IBS':2, 'TSI':2, 'GBR':2, 'IRN':2,
'CHB':3, 'JPT':3, 'KHV':3, 'CDX':3, 'CHS':3,
'BEB':4, 'STU':4, 'ITU':4, 'PJL':4, 'GIH':4}
pop = modelpops[selpop]
#colorDict = {1:'#FFB933', 2:'#0EBFF0', 3:'#ADCD00', 4:'#8B08B0'} #1000 Genomes group color scheme
colorDict = {1:'#cec627', 2:'#0EBFF0', 3:'#65ff00', 4:'#8B08B0'} #make it pop-!
f, axarr = plt.subplots(numChr, 1, sharex = True, sharey=True, dpi=dpi, figsize=(7, 10))
plt.suptitle(titlestring, fontsize=10)
plt.xlabel('position')
plt.ylabel('cms_gw normed score')
all_emp_pos, all_emp_scores = [], []
for chrom in range(1,numChr +1):
emp_cms_filename = get_emp_cms_file(selpop, chrom, normed=True, suffix=suffix, basedir=basedir)
print('loading chr ' + str(chrom) + ": " + emp_cms_filename)
if not os.path.isfile(emp_cms_filename):
print("missing: " + emp_cms_filename)
break
physpos, genpos, seldaf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(emp_cms_filename)
iax = chrom-1
ax = axarr[iax]
#ax.grid()
plot_data = eval(plotscore)
plotManhattan_extended(ax, plot_data, physpos, chrom)
all_emp_pos.append(physpos)
all_emp_scores.append(plot_data)
################################
## HILITE SIGNIFICANT REGIONS ##
################################
if args.regionsfile is not None:
regionchrs, regionstarts, regionends = load_regions(args.regionsfile)
print('loaded ' + str(len(regionchrs)) + ' significant regions from ' + args.regionsfile)
for iregion in range(len(regionchrs)):
regionchr, regionstart, regionend = regionchrs[iregion], regionstarts[iregion], regionends[iregion]
this_chrom = int(regionchr.strip('chr'))
ichrom = this_chrom-1
chrompos, chromscores = all_emp_pos[ichrom], all_emp_scores[ichrom]
zipped = zip(chrompos, chromscores)
plotpos, plotvals = [], []
for locus in zipped:
if locus[0] >= regionstart:
plotpos.append(locus[0])
plotvals.append(locus[1])
if locus[0] > regionend:
break
axarr[ichrom].plot(plotpos, plotvals, color=colorDict[pop], markersize=1)
if args.percentile is not None:
percentile = float(args.percentile)
print('plotting data with heuristic cutoff for ' + str(percentile) + " percentile...")
flat_emp_scores = [item for sublist in all_emp_scores for item in sublist if not np.isnan(item)]
score_cutoff = float(np.percentile(flat_emp_scores, percentile))
print("score cutoff: " + str(score_cutoff))
for chrom in range(1,numChr +1):
iax = chrom-1
ax = axarr[iax]
maximumVal = ax.get_xlim()[1]
xpoints = np.array([0, maximumVal])
ypoints = np.array([score_cutoff, score_cutoff])
ax.plot(xpoints, ypoints ,linestyle = "dotted", color="red", markersize=.3)
#get empirical scores and positions for pass threshhold and plot them as above with color
these_scores, these_pos = all_emp_scores[iax], all_emp_pos[iax]
zipped = zip(these_scores, these_pos)
significant = [item for item in zipped if item[0] >= score_cutoff]
signif_vals = [item[0] for item in significant]
signif_pos = [item[1] for item in significant]
ax.plot(signif_pos, signif_vals, color=colorDict[pop], linestyle='None', marker=".", markersize=.3)#, markersize=1)
plt.savefig(savename)
print('saved to: ' + savename)
return
def execute_regionviz(args):
''' visualize component and composite scores for a region '''
savefilename = args.savefilename
cmsfilename = args.cmsInfile
if os.path.isfile(cmsfilename):
print('loading from... ' + cmsfilename)
physpos, genpos, daf, ihs_normed, delihh_normed, nsl_normed, xpehh_normed, fst, deldaf, cms_unnormed, cms_normed = read_cms_repfile(cmsfilename) #need to make this flexible to regional input vs gw. (vs. likes)
causal_index = -1
if args.hilitePos is not None:
if args.hilitePos in physpos:
causal_index = physpos.index(args.hilitePos)
f, (ax1, ax2, ax3, ax4, ax5, ax6, ax7) = plt.subplots(7, sharex = True)
quick_plot(ax1, physpos, ihs_normed, "ihs_normed", causal_index)
quick_plot(ax2, physpos, delihh_normed, "delihh_normed", causal_index)
quick_plot(ax3, physpos, nsl_normed, "nsl_normed", causal_index)
quick_plot(ax4, physpos, xpehh_normed, "xpehh_normed", causal_index)
quick_plot(ax5, physpos, fst, "fst", causal_index)
quick_plot(ax6, physpos, deldaf, "deldaf", causal_index)
quick_plot(ax7, physpos, cms_unnormed, "cms", causal_index)
plt.savefig(savefilename)
print("plotted to " + savefilename)
plt.close()
return