def test_parse_mutation():
mut = ReachProcessor._parse_mutation('V600E')
assert mut.residue_from == 'V'
assert mut.position == '600'
assert mut.residue_to == 'E'
mut = ReachProcessor._parse_mutation('Leu174Arg')
assert mut.residue_from == 'L'
assert mut.position == '174'
assert mut.residue_to == 'R'
mut = ReachProcessor._parse_mutation('val34leu')
assert mut.residue_from == 'V'
assert mut.position == '34'
assert mut.residue_to == 'L'
def test_parse_mutation():
mut = ReachProcessor._parse_mutation('V600E')
assert mut.residue_from == 'V'
assert mut.position == '600'
assert mut.residue_to == 'E'
mut = ReachProcessor._parse_mutation('Leu174Arg')
assert mut.residue_from == 'L'
assert mut.position == '174'
assert mut.residue_to == 'R'
mut = ReachProcessor._parse_mutation('val34leu')
assert mut.residue_from == 'V'
assert mut.position == '34'
assert mut.residue_to == 'L'
def get_sites(self):
"""Parse the site-text string and return a list of sites.
Returns
-------
sites : list[Site]
A list of position-residue pairs corresponding to the site-text
"""
st = self.site_text
suffixes = [' residue', ' residues', ',', '/']
for suffix in suffixes:
if st.endswith(suffix):
st = st[:-len(suffix)]
assert(not st.endswith(','))
# Strip parentheses
st = st.replace('(', '')
st = st.replace(')', '')
st = st.replace(' or ', ' and ') # Treat end and or the same
sites = []
parts = st.split(' and ')
for part in parts:
if part.endswith(','):
part = part[:-1]
if len(part.strip()) > 0:
sites.extend(ReachProcessor._parse_site_text(part.strip()))
return sites
def get_sites(self):
"""Parse the site-text string and return a list of sites.
Returns
-------
sites : list[Site]
A list of position-residue pairs corresponding to the site-text
"""
st = self.site_text
suffixes = [' residue', ' residues', ',', '/']
for suffix in suffixes:
if st.endswith(suffix):
st = st[:-len(suffix)]
assert (not st.endswith(','))
# Strip parentheses
st = st.replace('(', '')
st = st.replace(')', '')
st = st.replace(' or ', ' and ') # Treat end and or the same
sites = []
parts = st.split(' and ')
for part in parts:
if part.endswith(','):
part = part[:-1]
if len(part.strip()) > 0:
sites.extend(ReachProcessor._parse_site_text(part.strip()))
return sites
def test_get_db_refs_up_non_human():
entity_term = {
'text': 'MYC',
'xrefs': [{'namespace': 'uniprot', 'id': 'Q9MZT7',
'object-type': 'db-reference'}]
}
db_refs = ReachProcessor._get_db_refs(entity_term)
assert db_refs == {'UP': 'Q9MZT7', 'TEXT': 'MYC'}, db_refs
def test_parse_site_text_number_first_space():
t = '293 T'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'T'
assert site == '293'
assert unicode_strs((residue, site))
def test_parse_site_text_number():
t = '135'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue is None
assert site == '135'
assert unicode_strs(site)
def test_parse_site_text_number():
t = '135'
sites = ReachProcessor._parse_site_text(t)
assert(len(sites) == 1)
residue, site = sites[0]
assert(residue is None)
assert(site == '135')
assert(unicode_strs(site))
def test_parse_site_text_number():
t = '135'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue is None
assert site == '135'
assert unicode_strs(site)
def test_parse_site_text_number_first():
t = '293T'
sites = ReachProcessor._parse_site_text(t)
assert(len(sites) == 1)
residue, site = sites[0]
assert(residue == 'T')
assert(site == '293')
assert(unicode_strs((residue, site)))
def test_parse_site_text_other_aa():
t = 'A431'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'A'
assert site == '431'
assert unicode_strs((residue, site))
def test_parse_site_residue_only():
text = ['serine residue', 'serine', 'a serine site', 's', 'ser']
assert unicode_strs(text)
for t in text:
residue, site = ReachProcessor._parse_site_text(t)
assert unicode_strs((residue, site))
assert (residue == 'S')
assert (site is None)
def test_parse_site_text_other_aa():
t = 'A431'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'A'
assert site == '431'
assert unicode_strs((residue, site))
def test_parse_site_text_number_first_space():
t = '293 T'
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'T'
assert site == '293'
assert unicode_strs((residue, site))
def test_get_db_refs_up_human():
entity_term = {
'text': 'Ikaros',
'xrefs': [{'namespace': 'uniprot', 'id': 'Q13422',
'object-type': 'db-reference'}]
}
db_refs = ReachProcessor._get_db_refs(entity_term)
assert db_refs == {'UP': 'Q13422', 'HGNC': '13176',
'TEXT': 'Ikaros', 'EGID': '10320'}, db_refs
def test_get_db_refs_up_non_human():
entity_term = {
'text': 'MYC',
'xrefs': [{'namespace': 'uniprot', 'id': 'Q9MZT7',
'object-type': 'db-reference'}]
}
name, db_refs = ReachProcessor._get_db_refs(entity_term)
assert name == 'MYC', name
assert db_refs == {'UP': 'Q9MZT7', 'TEXT': 'MYC'}, db_refs
def test_get_db_refs_up_human():
entity_term = {
'text': 'Ikaros',
'xrefs': [{'namespace': 'uniprot', 'id': 'Q13422',
'object-type': 'db-reference'}]
}
name, db_refs = ReachProcessor._get_db_refs(entity_term)
assert name == 'IKZF1', name
assert db_refs == {'UP': 'Q13422', 'HGNC': '13176',
'TEXT': 'Ikaros'}, db_refs
def test_parse_site_text():
text = [
'threonine 185', 'thr 185', 'thr-185', 'threonine residue 185', 'T185'
]
assert unicode_strs(text)
for t in text:
residue, site = ReachProcessor._parse_site_text(t)
assert (residue == 'T')
assert (site == '185')
assert unicode_strs((residue, site))
def test_parse_site_residue_only():
text = ['serine residue', 'serine', 'a serine site', 's', 'ser']
assert unicode_strs(text)
for t in text:
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert unicode_strs((residue, site))
assert residue == 'S'
assert site is None
def test_parse_site_residue_only():
text = ['serine residue', 'serine', 'a serine site', 's', 'ser']
assert unicode_strs(text)
for t in text:
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert unicode_strs((residue, site))
assert residue == 'S'
assert site is None
def test_parse_site_text():
text = ['threonine 185', 'thr 185', 'thr-185',
'threonine residue 185', 'T185']
assert unicode_strs(text)
for t in text:
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'T'
assert site == '185'
assert unicode_strs((residue, site))
def test_parse_site_text():
text = ['threonine 185', 'thr 185', 'thr-185',
'threonine residue 185', 'T185']
assert unicode_strs(text)
for t in text:
sites = ReachProcessor._parse_site_text(t)
assert len(sites) == 1
residue, site = sites[0]
assert residue == 'T'
assert site == '185'
assert unicode_strs((residue, site))
def test_valid_name():
assert (ReachProcessor._get_valid_name('') == '')
assert (ReachProcessor._get_valid_name('a') == 'a')
assert (ReachProcessor._get_valid_name('Name123') == 'Name123')
assert (ReachProcessor._get_valid_name('<>#~!,./][;-') == '____________')
assert (ReachProcessor._get_valid_name('PI3 Kinase') == 'PI3_Kinase')
assert (ReachProcessor._get_valid_name('14-3-3') == 'p14_3_3')
def test_parse_site_multiple():
sites = ReachProcessor._parse_site_text('638/641')
assert len(sites) == 2
assert sites[0][0] is None
assert sites[0][1] == '638'
assert sites[1][0] is None
assert sites[1][1] == '641'
sites = ReachProcessor._parse_site_text('992,1068')
assert len(sites) == 2
assert sites[0][0] is None
assert sites[0][1] == '992'
assert sites[1][0] is None
assert sites[1][1] == '1068'
sites = ReachProcessor._parse_site_text('Y1221/1222')
assert len(sites) == 2
assert sites[0][0] == 'Y'
assert sites[0][1] == '1221'
assert sites[1][0] == 'Y'
assert sites[1][1] == '1222'
sites = ReachProcessor._parse_site_text('Tyr-577/576')
assert len(sites) == 2
assert sites[0][0] == 'Y'
assert sites[0][1] == '577'
assert sites[1][0] == 'Y'
assert sites[1][1] == '576'
sites = ReachProcessor._parse_site_text('S199/S202/T205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] == 'S'
assert sites[1][1] == '202'
assert sites[2][0] == 'T'
assert sites[2][1] == '205'
sites = ReachProcessor._parse_site_text('S199/202/T205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] is None
assert sites[1][1] == '202'
assert sites[2][0] == 'T'
assert sites[2][1] == '205'
sites = ReachProcessor._parse_site_text('S199/202/205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] == 'S'
assert sites[1][1] == '202'
assert sites[2][0] == 'S'
assert sites[2][1] == '205'
def test_parse_site_multiple():
sites = ReachProcessor._parse_site_text('638/641')
assert len(sites) == 2
assert sites[0][0] is None
assert sites[0][1] == '638'
assert sites[1][0] is None
assert sites[1][1] == '641'
sites = ReachProcessor._parse_site_text('992,1068')
assert len(sites) == 2
assert sites[0][0] is None
assert sites[0][1] == '992'
assert sites[1][0] is None
assert sites[1][1] == '1068'
sites = ReachProcessor._parse_site_text('Y1221/1222')
assert len(sites) == 2
assert sites[0][0] == 'Y'
assert sites[0][1] == '1221'
assert sites[1][0] == 'Y'
assert sites[1][1] == '1222'
sites = ReachProcessor._parse_site_text('Tyr-577/576')
assert len(sites) == 2
assert sites[0][0] == 'Y'
assert sites[0][1] == '577'
assert sites[1][0] == 'Y'
assert sites[1][1] == '576'
sites = ReachProcessor._parse_site_text('S199/S202/T205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] == 'S'
assert sites[1][1] == '202'
assert sites[2][0] == 'T'
assert sites[2][1] == '205'
sites = ReachProcessor._parse_site_text('S199/202/T205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] is None
assert sites[1][1] == '202'
assert sites[2][0] == 'T'
assert sites[2][1] == '205'
sites = ReachProcessor._parse_site_text('S199/202/205')
assert len(sites) == 3
assert sites[0][0] == 'S'
assert sites[0][1] == '199'
assert sites[1][0] == 'S'
assert sites[1][1] == '202'
assert sites[2][0] == 'S'
assert sites[2][1] == '205'
def test_parse_site_text_other_aa():
t = 'A431'
residue, site = ReachProcessor._parse_site_text(t)
assert (residue == 'A')
assert (site == '431')
assert (unicode_strs((residue, site)))
def test_parse_site_text_number_first_space():
t = '293 T'
residue, site = ReachProcessor._parse_site_text(t)
assert (residue == 'T')
assert (site == '293')
assert (unicode_strs((residue, site)))