def write_hier_up(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
mark_dct = {'GO:0000001':'a', 'GO:0000003':'b', 'GO:0000006':'c', 'GO:0000008':'d'}
out.write('\nTEST MARK DICT: 01->03->06->08->09:\n')
objwr = WrHierGO(gosubdag, item_marks=mark_dct, sortby=lambda o: o.item_id)
gos_printed = objwr.prt_hier_up(["GO:0000005"], out)
assert gos_printed == ['GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000003', 'GO:0000005']
def write_hier_mrk(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
mark_lst = [
'GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009'
]
objwr = WrHierGO(gosubdag, go_marks=mark_lst)
objwr.prt_hier_down("GO:0000001", out)
def test_write_hier_mrk(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
# > - GO:0000001 BP 9 L00 D00 top
# -- GO:0000002 BP 2 L01 D01 B
# --- GO:0000005 BP 1 L02 D02 c
# ---- GO:0000010 BP 0 L03 D04 ac
# > -- GO:0000003 BP 7 L01 D01 A
# --- GO:0000004 BP 2 L02 D02 b
# ---- GO:0000007 BP 1 L03 D03 b1
# > ----- GO:0000009 BP 0 L04 D04 ab
# --- GO:0000005 BP 1 L02 D02 c
# ---- GO:0000010 BP 0 L03 D04 ac
# > --- GO:0000006 BP 3 L02 D02 a
# > ---- GO:0000008 BP 2 L03 D03 a1
# > ----- GO:0000009 BP 0 L04 D04 ab
# ----- GO:0000010 BP 0 L03 D04 ac
mark_lst = [
'GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009'
]
out.write('\nTEST {NAME} MARK: 01->03->06->08->09:\n'.format(NAME="MINI"))
objwr = WrHierGO(gosubdag, item_marks=mark_lst, sortby=lambda o: o.item_id)
gos_prtd = objwr.prt_hier_down("GO:0000001", out)
assert gos_prtd == [
'GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000010', 'GO:0000003',
'GO:0000004', 'GO:0000007', 'GO:0000009', 'GO:0000005', 'GO:0000010',
'GO:0000006', 'GO:0000008', 'GO:0000009', 'GO:0000010'
]
def write_hier_mrk(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
mark_lst = [
'GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009'
]
out.write('\nTEST MARK: 01->03->06->08->09:\n')
objwr = WrHierGO(gosubdag, go_marks=mark_lst)
objwr.prt_hier_down("GO:0000001", out)
def write_hier_lim(gosubdag, out):
"""Limits hierarchy list to GO Terms specified by user."""
go_omit = ['GO:0000005', 'GO:0000010']
go_ids = [go_id for go_id in gosubdag.go2obj if go_id not in go_omit]
out.write('\nTEST OMIT: 05 and 10:\n')
objwr = WrHierGO(gosubdag, include_only=go_ids)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert not gos_printed.intersection(go_omit), "SHOULD NOT PRINT {GOs}".format(GOs=go_omit)
def prt_hier(self, prt=sys.stdout):
"""Write hierarchy below specfied GO IDs."""
objwr = WrHierGO(self.gosubdag, **self.kws)
assert self.goids, "NO VALID GO IDs WERE PROVIDED"
if 'up' not in objwr.usrset:
for goid in self.goids:
objwr.prt_hier_down(goid, prt)
else:
objwr.prt_hier_up(self.goids, prt)
def write_hier_mrk_lst(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
mark_lst = ['GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009']
out.write('\nTEST MARK LIST: 01->03->06->08->09:\n')
objwr = WrHierGO(gosubdag, item_marks=mark_lst, sortby=lambda o: o.item_id)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert gos_printed == ['GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000010',
'GO:0000003', 'GO:0000004', 'GO:0000007', 'GO:0000009',
'GO:0000005', 'GO:0000010', 'GO:0000006', 'GO:0000008',
'GO:0000009', 'GO:0000010']
def write_hier_lim(gosubdag, out):
"""Limits hierarchy list to GO Terms specified by user."""
go_omit = ['GO:0000005', 'GO:0000010']
go_ids = [go_id for go_id in gosubdag.go2obj if go_id not in go_omit]
out.write('\nTEST OMIT: 05 and 10:\n')
objwr = WrHierGO(gosubdag, include_only=go_ids, sortby=lambda o: o.item_id)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert not set(gos_printed).intersection(go_omit), "SHOULD NOT PRINT {GOs}".format(GOs=go_omit)
assert gos_printed == ['GO:0000001', 'GO:0000002', 'GO:0000003', 'GO:0000004',
'GO:0000007', 'GO:0000009', 'GO:0000006', 'GO:0000008',
'GO:0000009']
def extract_hier_all(gosubdag, out, root_term, go2geneids, prize_dictionary):
"""write_hier.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST EXTRACTION: Print all hierarchies:\n')
objwr = WrHierGO(gosubdag,
go2geneids=go2geneids,
prize_dictionary=prize_dictionary,
G=nx.DiGraph())
obj = objwr.ext_hier_down(root_term, out)
seeds = prize_dictionary.keys()
layer_i_nodes = linear_threshold(obj.G, seeds, steps=INFLUENCE_N_STEPS)
return (obj.vertices, obj.edges, obj.G, layer_i_nodes)
def write_hier_norep(gosubdag, out):
"""Shortens hierarchy report by only printing branches once.
Prints the 'entire hierarchy' of GO:0000005 the 1st time seen:
--- 1 GO:0000005 L-02 D-02
---- 0 GO:0000010 L-03 D-04
Prints just GO:0000005 (ommit child GO:10) the 2nd time seen:
=== 1 GO:0000005 L-02 D-02
'=' is used in hierarchy mark to indicate that the pathes
below the marked term have already been printed.
"""
# out.write('\nTEST ALL: Print branches just once:\n')
objwr = WrHierGO(gosubdag, concise=True)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert gos_printed == set(objwr.gosubdag.go2nt)
def test_write_hier_all(name, go_id, gosubdag, out):
"""test_optional_fields.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST {NAME} {GO}: Print all hierarchies:\n'.format(NAME=name,
GO=go_id))
#### dag.write_hier(go_id, out, num_child=True)
WrHierGO(gosubdag).prt_hier_down(go_id, out)
out.write("GOTerm: {}\n".format(gosubdag.go2obj[go_id].__repr__()))
def test_write_hier_mrk(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
mark_lst = ['GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009']
out.write('\nTEST {NAME} MARK: 01->03->06->08->09:\n'.format(NAME="MINI"))
# TBD: Add go_marks fnc to wr_hier
#### dag.write_hier("GO:0000001", out, go_marks=mark_lst)
WrHierGO(gosubdag).prt_hier_down("GO:0000001", out)
def test_write_hier_bp_mf_cc():
"""Test that write hierarchy writes all: BP, MF, CC"""
fin_anno = os.path.join(REPO, 'gene2go')
fin_dag = os.path.join(REPO, "go-basic.obo")
_dnld_anno(fin_anno)
#godag = get_godag(os.path.join(REPO, 'go-basic.obo'), loading_bar=None)
print('\nTEST STORING ONLY ONE SPECIES')
#### obj = Gene2GoReader(fin_anno)
godag = get_godag(fin_dag)
gene2gos = read_annotations(namespace='ALL')
tcntobj = TermCounts(godag, gene2gos) if gene2gos else None
gosubdag = GoSubDag(godag.keys(),
godag,
relationships=False,
tcntobj=tcntobj,
children=True,
prt=sys.stdout)
objwr = WrHierGO(gosubdag)
# 2020 11:
# 594,748 GO lines under GO:0008150
# 23,199 GO lines under GO:0003674
# 6,259 GO lines under GO:0005575
# 624,206 items WROTE: tmp_test_wr_hier_BP_MF_CC.txt
assert len(_wr_hier(['BP', 'MF', 'CC'], gosubdag.go2nt, objwr)) > 600000
assert len(_wr_hier([
'BP',
], gosubdag.go2nt, objwr)) > 500000
assert len(_wr_hier([
'MF',
], gosubdag.go2nt, objwr)) > 20000
assert len(_wr_hier([
'CC',
], gosubdag.go2nt, objwr)) > 5000
def test_write_hier_lim(gosubdag, out):
"""Limits hierarchy list to GO Terms specified by user."""
go_omit = ['GO:0000005', 'GO:0000010']
go_ids = [go_id for go_id in gosubdag.go2nt.keys() if go_id not in go_omit]
out.write('\nTEST {NAME} OMIT: 05->10:\n'.format(NAME="MINI"))
# TBD: Add include_only fnc to wr_hier
#### dag.write_hier("GO:0000001", out, include_only=go_ids)
WrHierGO(gosubdag).prt_hier_down("GO:0000001", out)
def test_write_hier_lim(gosubdag, out):
"""Limits hierarchy list to GO Terms specified by user."""
# - GO:0000001 BP 9 L00 D00 top
# -- GO:0000002 BP 2 L01 D01 B
# -- GO:0000003 BP 7 L01 D01 A
# --- GO:0000004 BP 2 L02 D02 b
# ---- GO:0000007 BP 1 L03 D03 b1
# ----- GO:0000009 BP 0 L04 D04 ab
# --- GO:0000006 BP 3 L02 D02 a
# ---- GO:0000008 BP 2 L03 D03 a1
# ----- GO:0000009 BP 0 L04 D04 ab
go_omit = ['GO:0000005']
go_ids = [go_id for go_id in gosubdag.go2nt.keys() if go_id not in go_omit]
out.write('\nTEST {NAME} OMIT: 05->10:\n'.format(NAME="MINI"))
objwr = WrHierGO(gosubdag, include_only=go_ids, sortby=lambda o: o.item_id)
gos_prtd = objwr.prt_hier_down("GO:0000001", out)
print(gos_prtd)
assert gos_prtd == ['GO:0000001', 'GO:0000002', 'GO:0000003', 'GO:0000004', 'GO:0000007',
'GO:0000009', 'GO:0000006', 'GO:0000008', 'GO:0000009', 'GO:0000010']
def test_write_hier_lim(gosubdag, out):
"""Limits hierarchy list to GO Terms specified by user."""
# - GO:0000001 BP 9 L00 D00 top
# -- GO:0000002 BP 2 L01 D01 B
# -- GO:0000003 BP 7 L01 D01 A
# --- GO:0000004 BP 2 L02 D02 b
# ---- GO:0000007 BP 1 L03 D03 b1
# ----- GO:0000009 BP 0 L04 D04 ab
# --- GO:0000006 BP 3 L02 D02 a
# ---- GO:0000008 BP 2 L03 D03 a1
# ----- GO:0000009 BP 0 L04 D04 ab
go_omit = ['GO:0000005']
go_ids = [go_id for go_id in gosubdag.go2nt.keys() if go_id not in go_omit]
out.write('\nTEST {NAME} OMIT: 05->10:\n'.format(NAME="MINI"))
objwr = WrHierGO(gosubdag, include_only=go_ids, sortby=lambda o: o.item_id)
gos_prtd = objwr.prt_hier_down("GO:0000001", out)
print(gos_prtd)
assert gos_prtd == [
'GO:0000001', 'GO:0000002', 'GO:0000003', 'GO:0000004', 'GO:0000007',
'GO:0000009', 'GO:0000006', 'GO:0000008', 'GO:0000009', 'GO:0000010'
]
def test_write_hier_mrk(gosubdag, out):
"""Print all paths, but mark GO Terms of interest. """
# > - GO:0000001 BP 9 L00 D00 top
# -- GO:0000002 BP 2 L01 D01 B
# --- GO:0000005 BP 1 L02 D02 c
# ---- GO:0000010 BP 0 L03 D04 ac
# > -- GO:0000003 BP 7 L01 D01 A
# --- GO:0000004 BP 2 L02 D02 b
# ---- GO:0000007 BP 1 L03 D03 b1
# > ----- GO:0000009 BP 0 L04 D04 ab
# --- GO:0000005 BP 1 L02 D02 c
# ---- GO:0000010 BP 0 L03 D04 ac
# > --- GO:0000006 BP 3 L02 D02 a
# > ---- GO:0000008 BP 2 L03 D03 a1
# > ----- GO:0000009 BP 0 L04 D04 ab
# ----- GO:0000010 BP 0 L03 D04 ac
mark_lst = ['GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000009']
out.write('\nTEST {NAME} MARK: 01->03->06->08->09:\n'.format(NAME="MINI"))
objwr = WrHierGO(gosubdag, item_marks=mark_lst, sortby=lambda o: o.item_id)
gos_prtd = objwr.prt_hier_down("GO:0000001", out)
assert gos_prtd == ['GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000010', 'GO:0000003',
'GO:0000004', 'GO:0000007', 'GO:0000009', 'GO:0000005', 'GO:0000010',
'GO:0000006', 'GO:0000008', 'GO:0000009', 'GO:0000010']
def write_hier_norep(gosubdag, out):
"""Shortens hierarchy report by only printing branches once.
Prints the 'entire hierarchy' of GO:0000005 the 1st time seen:
--- 1 GO:0000005 L-02 D-02
---- 0 GO:0000010 L-03 D-04
Prints just GO:0000005 (ommit child GO:10) the 2nd time seen:
=== 1 GO:0000005 L-02 D-02
'=' is used in hierarchy mark to indicate that the pathes
below the marked term have already been printed.
"""
out.write('\nTEST ALL: Print branches just once:\n')
objwr = WrHierGO(gosubdag, concise=True, sortby=lambda o: o.item_id)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert set(gos_printed) == set(objwr.gosubdag.go2nt)
assert gos_printed == ['GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000010',
'GO:0000003', 'GO:0000004', 'GO:0000007', 'GO:0000009',
'GO:0000005', 'GO:0000006', 'GO:0000008', 'GO:0000009',
'GO:0000010']
def prt_hier(self, prt=sys.stdout):
"""Write hierarchy below specfied GO IDs."""
objwr = WrHierGO(self.gosubdag, **self.kws)
assert self.goids, "NO VALID GO IDs WERE PROVIDED"
if 'up' not in objwr.usrset:
for goid in self.goids:
objwr.prt_hier_down(goid, prt)
else:
objwr.prt_hier_up(self.goids, prt)
def prt_hier(self, prt=sys.stdout):
"""Write hierarchy below specfied GO IDs."""
objwr = WrHierGO(self.gosubdag, **self.kws)
assert self.goids, "NO VALID GO IDs WERE PROVIDED"
# kws = {k:v for k, v in self.kws.items() if k in self.kws_dct_wr}
# objwr.write_hier_all(prt=prt, **kws)
# max_indent=None, num_child=None, short_prt=False):
if 'up' not in objwr.usrset:
for goid in self.goids:
objwr.prt_hier_down(goid, prt)
else:
objwr.prt_hier_up(self.goids, prt)
def test_write_hier_norep(name, go_id, gosubdag, out):
"""Shortens hierarchy REPOrt by only printing branches once.
Prints the 'entire hierarchy' of GO:0000005 the 1st time seen:
--- 1 GO:0000005 L-02 D-02
---- 0 GO:0000010 L-03 D-04
Prints just GO:0000005 (ommit child GO:10) the 2nd time seen:
=== 1 GO:0000005 L-02 D-02
'=' is used in hierarchy mark to indicate that the pathes
below the marked term have already been printed.
"""
out.write('\nTEST {NAME} {GO}: Print branches just once:\n'.format(
NAME=name, GO=go_id))
#### dag.write_hier(go_id, out, num_child=True, short_prt=True)
WrHierGO(gosubdag, concise=True).prt_hier_down(go_id, out)
out.write("GOTerm: {}\n".format(gosubdag.go2obj[go_id]))
def write_hier_all(gosubdag, out):
"""write_hier.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST ALL: Print all hierarchies:\n')
objwr = WrHierGO(gosubdag)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert set(gos_printed) == set(objwr.gosubdag.go2nt)
def extract_hier_all(gosubdag, out, root_term, go2geneids):
"""write_hier.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST EXTRACTION: Print all hierarchies:\n')
objwr = WrHierGO(gosubdag, go2geneids=go2geneids)
obj = objwr.ext_hier_down(root_term, out)
return (obj.vertices, obj.edges)
def write_hier_all(gosubdag, out, root_term):
"""write_hier.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST ALL: Print all hierarchies:\n')
objwr = WrHierGO(gosubdag)
gos_printed = objwr.prt_hier_down(root_term, out)
print len(gos_printed)
def write_hier_all(gosubdag, out):
"""write_hier.py: Prints the entire mini GO hierarchy, with counts of children."""
out.write('\nTEST ALL: Print all hierarchies:\n')
objwr = WrHierGO(gosubdag)
gos_printed = objwr.prt_hier_down("GO:0000001", out)
assert set(gos_printed) == set(objwr.gosubdag.go2nt)