def test_assc_stats(prt=sys.stdout):
"""Test association statistics."""
associations = [
('hsa', 'goa_human.gaf'), # human
('mus', 'mgi.gaf'), # mouse
('dme', 'fb.gaf')] # fly
godag = get_godag(os.path.join(REPO, "go-basic.obo"), loading_bar=None)
describe_go2obj(godag, prt)
obj = StatsDescribe('Assc', "{:6,}")
obj.prt_hdr(prt, "Assc.")
for org, assc_name in associations:
fin_assc = os.path.join(REPO, assc_name)
describe_assc(org, fin_assc, godag, obj, prt)
def test_assc_stats(prt=sys.stdout):
"""Test association statistics."""
associations = [
('hsa', 'goa_human.gaf'), # human
('mus', 'mgi.gaf'), # mouse
('dme', 'fb.gaf')
] # fly
godag = get_godag(os.path.join(REPO, "go-basic.obo"), loading_bar=None)
describe_go2obj(godag, prt)
obj = StatsDescribe('Assc', "{:6,}")
obj.prt_hdr(prt, "Assc.")
for org, assc_name in associations:
fin_assc = os.path.join(REPO, assc_name)
describe_assc(org, fin_assc, godag, obj, prt)
def test_statsdescribe():
"""Use StatsDescribe to create a markdown table.
fdr_bh
name | # fdr_bh | range of fdr_bh | 25th perc| median | 75th perc| mean | stddev
---------|----------|----------------------|----------|----------|----------|----------|---------
GOATOOLS | 59 | 1.87e-07 to 4.94e-02 | 2.72e-04 | 1.03e-02 | 3.04e-02 | 1.56e-02 | 1.82e-02
"""
#pylint: disable=no-member
nts_goids = [nt for nt in nts if nt.p_fdr_bh < 0.05]
fdr_vals = [nt.p_fdr_bh for nt in nts_goids]
statsobj = StatsDescribe("fdr_bh", fmtstr="{:>8.2e}")
statsobj.prt_hdr()
statsobj.prt_data("GOATOOLS", fdr_vals)
def test_statsdescribe():
"""Use StatsDescribe to create a markdown table.
fdr_bh
name | # fdr_bh | range of fdr_bh | 25th perc| median | 75th perc| mean | stddev
---------|----------|----------------------|----------|----------|----------|----------|---------
GOATOOLS | 59 | 1.87e-07 to 4.94e-02 | 2.72e-04 | 1.03e-02 | 3.04e-02 | 1.56e-02 | 1.82e-02
"""
#pylint: disable=no-member
# Somehow goea_results contains fields of empty string, which we can check with:
# print([(nt.GO, nt.p_fdr_bh) for nt in goea_results])
nts_goids = [nt for nt in goea_results if nt.p_fdr_bh != '' and nt.p_fdr_bh < 0.05]
fdr_vals = [nt.p_fdr_bh for nt in nts_goids]
statsobj = StatsDescribe("fdr_bh", fmtstr="{:>8.2e}")
statsobj.prt_hdr()
statsobj.prt_data("GOATOOLS", fdr_vals)
def describe_go2obj(go2obj, prt):
"""Describe distribution of parent and child GO term counts."""
# Related GO | # GO | range | 25th | median | 75th | mean | stddev
# -----------|-------|----------|------|--------|------|------|-------
# Parents | 44961 | 0 to 8 | 1 | 1 | 2 | 2 | 1
# Children | 17597 | 1 to 480 | 1 | 2 | 4 | 4 | 10
cnts_all = [(len(o.children), len(o.parents)) for go, o in go2obj.items()
if go == o.id]
cnts_c, cnts_p = zip(*cnts_all)
cnts_c = [n for n in cnts_c
if n != 0] # Remove leaf-level counts from reported stats
cnts_p = [n for n in cnts_p
if n != 0] # Remove top-level counts from reported stats
obj = StatsDescribe('GO', "{:6,}")
obj.prt_hdr(prt, "Related GO")
obj.prt_data("Parents", cnts_p, prt)
obj.prt_data("Children", cnts_c, prt)
def describe_go2obj(go2obj, prt):
"""Describe distribution of parent and child GO term counts."""
# Related GO | # GO | range | 25th | median | 75th | mean | stddev
# -----------|-------|----------|------|--------|------|------|-------
# Parents | 44961 | 0 to 8 | 1 | 1 | 2 | 2 | 1
# Children | 17597 | 1 to 480 | 1 | 2 | 4 | 4 | 10
cnts_all = [(len(o.children), len(o.parents)) for go, o in go2obj.items() if go == o.id]
cnts_c, cnts_p = zip(*cnts_all)
cnts_c = [n for n in cnts_c if n != 0] # Remove leaf-level counts from reported stats
cnts_p = [n for n in cnts_p if n != 0] # Remove top-level counts from reported stats
obj = StatsDescribe('GO', "{:6,}")
obj.prt_hdr(prt, "Related GO")
obj.prt_data("Parents", cnts_p, prt)
obj.prt_data("Children", cnts_c, prt)
def prt_experiments_stats(self, prt=sys.stdout, attrs=None, genes_goids='genes'):
"""Print stats for user-specified data in experiment sets."""
if attrs is None:
attrs = ["fdr_actual", "frr_actual", "num_Type_I", "num_Type_II", "num_correct"]
hdrexps = "Nul(% max) #pval #tests" # Header for col0, the description of the statistic
namefmt = "{PERCNULL:3}% {EXP_ALPHA:5.3f} {QTY:5}"
for attrname in attrs:
prt.write("\n{ATTR} statistics:\n".format(ATTR=attrname))
objstat = StatsDescribe("exps", "{:10.2f}" if attrname[:3] == "num" else "{:6.4f}")
objstat.prt_hdr(prt, hdrexps)
for experiment_set in self.expsets: # ExperimentSet
expname = experiment_set.get_desc(namefmt)
means = experiment_set.get_means(attrname, genes_goids)
objstat.prt_data(expname, means, prt)
def test_statsdescribe():
"""Use StatsDescribe to create a markdown table.
fdr_bh
name | # fdr_bh | range of fdr_bh | 25th perc| median | 75th perc| mean | stddev
---------|----------|----------------------|----------|----------|----------|----------|---------
GOATOOLS | 59 | 1.87e-07 to 4.94e-02 | 2.72e-04 | 1.03e-02 | 3.04e-02 | 1.56e-02 | 1.82e-02
"""
#pylint: disable=no-member
nts_goids = [nt for nt in nts if nt.p_fdr_bh < 0.05]
fdr_vals = [nt.p_fdr_bh for nt in nts_goids]
statsobj = StatsDescribe("fdr_bh", fmtstr="{:>8.2e}")
statsobj.prt_hdr()
statsobj.prt_data("GOATOOLS", fdr_vals)
def test_statsdescribe():
"""Use StatsDescribe to create a markdown table.
fdr_bh
name | # fdr_bh | range of fdr_bh | 25th perc| median | 75th perc| mean | stddev
---------|----------|----------------------|----------|----------|----------|----------|---------
GOATOOLS | 59 | 1.87e-07 to 4.94e-02 | 2.72e-04 | 1.03e-02 | 3.04e-02 | 1.56e-02 | 1.82e-02
"""
#pylint: disable=no-member
# Somehow nts contains fields of empty string, which we can check with:
# print([(nt.GO, nt.p_fdr_bh) for nt in nts])
nts_goids = [nt for nt in nts if nt.p_fdr_bh != '' and nt.p_fdr_bh < 0.05]
fdr_vals = [nt.p_fdr_bh for nt in nts_goids]
statsobj = StatsDescribe("fdr_bh", fmtstr="{:>8.2e}")
statsobj.prt_hdr()
statsobj.prt_data("GOATOOLS", fdr_vals)
def main(prt=sys.stdout):
"""Statistics for the protein-coding mouse gene association."""
godag = get_godag()
params = {
'association_file': os.path.join(REPO, 'gene_association.mgi'),
'genes_population': ensm2nt.keys()
} # Population genes
objassc = DataAssc(params, godag)
# Statistics for number of genes per GO in the mouse association for protein-coding genes
go2numgenes = {go: len(genes) for go, genes in objassc.go2genes.items()}
objdesc = StatsDescribe("GOs", "{:>5.0f}")
objdesc.prt_hdr(prt, name="\nname ")
objdesc.prt_data("# genes/GO", go2numgenes.values(), prt)
# Statistics for number of GOs per gene in the mouse association for protein-coding genes
gene2numgos = {
gene: len(gos)
for gene, gos in objassc.objassc_all.assc_geneid2gos.items()
}
objdesc = StatsDescribe("genes", "{:>5.0f}")
objdesc.prt_hdr(prt, name="\nname ")
objdesc.prt_data("# GOs/gene", gene2numgos.values(), prt)
# Percentage of Ensembl mouse genes covered by GO annotations
num_pc = len(params['genes_population'])
num_assc = len(objassc.objassc_all.assc_geneid2gos)
prt.write(
"{PERC:2.0f}% of {A} of {P} Mouse protein-coding genes are annotated by GO IDs.\n"
.format(PERC=100.0 * num_assc / num_pc, P=num_pc, A=num_assc))