def model_makeplots(indir, outdir):
objs = lcommon.load_pickledir(indir)
if not objs:
return
rcommon.union_models(objs)
for attr in objs[0].get_attrs_depth1():
outbase = os.path.join(outdir, attr)
k2ydata = model_get_ydata(objs, attr)
gene_attrs = ['v', 'd', 'j']
if attr in gene_attrs:
xlabels = lcommon.sort_by_gene_number(k2ydata.keys())
rcmp.diff_plot(k2ydata, outbase, xlabels, attr.upper(),
ylabel='Frequency')
else:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(), xmin=-0.5,
xmax=30.5, ylabel='Frequency')
for attr in objs[0].get_attrs_depth2():
k_k2ydata = model_get_ydata2(objs, attr)
for k, k2ydata in k_k2ydata.iteritems():
if not isinstance(k, str):
k = str(k)
outbase = os.path.join(outdir, "%s_%s" % (attr, k))
if attr in ['v2del', 'j2del', 'd2del']:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(), xmin=-0.5,
xmax=16.5, ylabel='Frequency')
else:
rcmp.diff_plot(k2ydata, outbase, label=attr.upper(),
ylabel='Frequency')
def model_diff(indir1, indir2, outdir):
# indir1 contains sam1_model.pickle, sam2_model.pickle ...
# indir2: similarly to indir1
name1 = os.path.basename(indir1.rstrip("/"))
name2 = os.path.basename(indir2.rstrip("/"))
objs1 = lcommon.load_pickledir(indir1)
objs2 = lcommon.load_pickledir(indir2)
rcommon.union_models(objs1 + objs2)
# statistic for each attr for differential usage:
statsdir = os.path.join(outdir, "stat_tests")
system("mkdir -p %s" % statsdir)
model_diff_signi_test(objs1, objs2, statsdir)
# if True: # HACK
# return
med1 = rcommon.model_get_median(objs1)
# med2 = model_get_median(objs2)
for attr in med1.get_attrs_depth1():
outbase1 = os.path.join(outdir, "%s_%s_cmp2_%s" % (attr, name2, name1))
k2diffs1 = diff_sample_model(med1, objs2, attr)
gene_attrs = ["v", "d", "j", "dj"]
if attr in gene_attrs:
if attr == "dj":
xlabels = sorted(k2diffs1.keys())
else:
xlabels = lcommon.sort_by_gene_number(k2diffs1.keys())
diff_plot(k2diffs1, outbase1, xlabels, attr.upper(), ymin=-0.1, ymax=0.1)
else:
diff_plot(k2diffs1, outbase1, label=attr.upper(), xmin=-0.5, xmax=30.5, ymin=-0.02, ymax=0.05)
# outbase2 = os.path.join(outdir, "%s_%s_cmp2_%s" % (attr, name1, name2))
# k2diffs2 = diff_sample_model(med2, objs1, attr)
# diff_plot(k2diffs2, outbase2)
for attr in med1.get_attrs_depth2():
k_k2diffs1 = diff_sample_model2(med1, objs2, attr)
for k, k2diffs1 in k_k2diffs1.iteritems():
if not isinstance(k, str):
k = str(k)
outbase1 = os.path.join(outdir, "%s_%s_%s_cmp2_%s" % (attr, k, name2, name1))
if attr in ["v2del", "j2del", "d2del"]:
diff_plot(k2diffs1, outbase1, label=attr.upper(), xmin=-0.5, xmax=16.5)
else:
diff_plot(k2diffs1, outbase1, label=attr.upper())
