def process_paper(model_name, pmid):
json_path = os.path.join(model_path, model_name, 'jsons',
'PMID%s.json' % pmid)
if pmid.startswith('api') or pmid.startswith('PMID'):
logger.warning('Invalid PMID: %s' % pmid)
# If the paper has been read, use the json output file
if os.path.exists(json_path):
rp = reach.process_json_file(json_path, citation=pmid)
txt_format = 'existing_json'
# If the paper has not been read, download the text and read
else:
txt, txt_format = get_full_text(pmid, 'pmid')
if txt_format == 'pmc_oa_xml':
rp = reach.process_nxml_str(txt, citation=pmid, offline=True)
if os.path.exists('reach_output.json'):
shutil.move('reach_output.json', json_path)
elif txt_format == 'elsevier_xml':
# Extract the raw text from the Elsevier XML
txt = elsevier_client.extract_text(txt)
rp = reach.process_text(txt, citation=pmid, offline=True)
if os.path.exists('reach_output.json'):
shutil.move('reach_output.json', json_path)
elif txt_format == 'abstract':
rp = reach.process_text(txt, citation=pmid, offline=True)
if os.path.exists('reach_output.json'):
shutil.move('reach_output.json', json_path)
else:
rp = None
if rp is not None:
check_pmids(rp.statements)
return rp, txt_format
def test_phosphorylate():
for offline in offline_modes:
rp = reach.process_text('MEK1 phosphorylates ERK2.', offline=offline)
assert (len(rp.statements) == 1)
s = rp.statements[0]
assert (s.enz.name == 'MAP2K1')
assert (s.sub.name == 'MAPK1')
assert unicode_strs(rp.statements)
def test_activate():
for offline in offline_modes:
rp = reach.process_text('HRAS activates BRAF.', offline=offline)
assert (len(rp.statements) == 1)
s = rp.statements[0]
assert (s.subj.name == 'HRAS')
assert (s.obj.name == 'BRAF')
assert unicode_strs(rp.statements)
def test_activate():
for offline in offline_modes:
rp = reach.process_text('HRAS activates BRAF.', offline=offline)
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert (s.subj.name == 'HRAS')
assert (s.obj.name == 'BRAF')
assert unicode_strs(rp.statements)
def test_phosphorylate():
for offline in offline_modes:
rp = reach.process_text('MEK1 phosphorylates ERK2.', offline=offline)
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert (s.enz.name == 'MAP2K1')
assert (s.sub.name == 'MAPK1')
assert unicode_strs(rp.statements)
def test_regulate_amount():
for offline in offline_modes:
rp = reach.process_text('ERK increases the transcription of DUSP.',
offline=offline)
assert (len(rp.statements) == 1)
s = rp.statements[0]
assert (isinstance(s, IncreaseAmount))
assert (s.subj.name == 'ERK')
assert (s.obj.name == 'DUSP')
assert unicode_strs(rp.statements)
rp = reach.process_text('ERK decreases the amount of DUSP.',
offline=offline)
assert (len(rp.statements) == 1)
s = rp.statements[0]
assert (isinstance(s, DecreaseAmount))
assert (s.subj.name == 'ERK')
assert (s.obj.name == 'DUSP')
assert unicode_strs(rp.statements)
def test_be_grounding():
for offline in offline_modes:
rp = reach.process_text('MEK activates ERK.', offline=offline)
assert (len(rp.statements) == 1)
assert unicode_strs(rp.statements)
if offline == True:
st = rp.statements[0]
assert (st.subj.db_refs.get('BE') == 'MEK')
assert (st.obj.db_refs.get('BE') == 'ERK')
def test_mutation():
rp = reach.process_text('BRAF(V600E) phosphorylates MEK.')
assert(len(rp.statements) == 1)
braf = rp.statements[0].enz
assert(braf.name == 'BRAF')
assert(len(braf.mutations) == 1)
assert(braf.mutations[0].position == '600')
assert(braf.mutations[0].residue_from == 'V')
assert(braf.mutations[0].residue_to == 'E')
def test_be_grounding():
for offline in offline_modes:
rp = reach.process_text('MEK activates ERK.', offline=offline)
assert(len(rp.statements) == 1)
assert unicode_strs(rp.statements)
if offline == True:
st = rp.statements[0]
assert(st.subj.db_refs.get('BE') == 'MEK')
assert(st.obj.db_refs.get('BE') == 'ERK')
def process_reach(txt):
print('Using REACH')
ts = time.time()
rp = reach.process_text(txt, offline=False)
for s in rp.statements:
print('%s\t%s' % (s, s.evidence[0].text))
te = time.time()
print('Time taken: %.2fs' % (te - ts))
return rp.statements
def test_regulate_amount():
for offline in offline_modes:
rp = reach.process_text('ERK increases the transcription of DUSP.',
offline=offline)
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert(isinstance(s, IncreaseAmount))
assert (s.subj.name == 'ERK')
assert (s.obj.name == 'DUSP')
assert unicode_strs(rp.statements)
rp = reach.process_text('ERK decreases the amount of DUSP.',
offline=offline)
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert(isinstance(s, DecreaseAmount))
assert (s.subj.name == 'ERK')
assert (s.obj.name == 'DUSP')
assert unicode_strs(rp.statements)
def test_mutation():
for offline in offline_modes:
rp = reach.process_text('BRAF(V600E) phosphorylates MEK.',
offline=offline)
assert (len(rp.statements) == 1)
braf = rp.statements[0].enz
assert (braf.name == 'BRAF')
assert (len(braf.mutations) == 1)
assert (braf.mutations[0].position == '600')
assert (braf.mutations[0].residue_from == 'V')
assert (braf.mutations[0].residue_to == 'E')
assert unicode_strs(rp.statements)
def test_mutation():
for offline in offline_modes:
rp = reach.process_text('BRAF(V600E) phosphorylates MEK.',
offline=offline)
assert(len(rp.statements) == 1)
braf = rp.statements[0].enz
assert(braf.name == 'BRAF')
assert(len(braf.mutations) == 1)
assert(braf.mutations[0].position == '600')
assert(braf.mutations[0].residue_from == 'V')
assert(braf.mutations[0].residue_to == 'E')
assert unicode_strs(rp.statements)
def reach_process_text():
"""Process text with REACH and return INDRA Statements."""
response = request.body.read().decode('utf-8')
body = json.loads(response)
text = body.get('text')
rp = reach.process_text(text)
if rp and rp.statements:
stmts = stmts_to_json(rp.statements)
res = {'statements': stmts}
return res
else:
res = {'statements': []}
return res
def test_hgnc_from_up():
for offline in offline_modes:
rp = reach.process_text('MEK1 phosphorylates ERK2.', offline=offline)
assert len(rp.statements) == 1
st = rp.statements[0]
(map2k1, mapk1) = st.agent_list()
assert map2k1.name == 'MAP2K1'
assert map2k1.db_refs['HGNC'] == '6840'
assert map2k1.db_refs['UP'] == 'Q02750'
assert mapk1.name == 'MAPK1'
assert mapk1.db_refs['HGNC'] == '6871'
assert mapk1.db_refs['UP'] == 'P28482'
assert unicode_strs(rp.statements)
def test_multiple_enzymes():
for offline in offline_modes:
rp = reach.process_text('MEK1 and MEK2 phosphorylate ERK1.',
offline=offline)
assert(len(rp.statements) == 2)
s = rp.statements[0]
if s.enz.name == 'MAP2K1':
assert(rp.statements[1].enz.name == 'MAP2K2')
else:
assert(rp.statements[1].enz.name == 'MAP2K1')
assert (s.sub.name == 'MAPK3')
s = rp.statements[1]
assert (s.sub.name == 'MAPK3')
assert unicode_strs(rp.statements)
def test_hgnc_from_up():
for offline in offline_modes:
rp = reach.process_text('MEK1 phosphorylates ERK2.',
offline=offline)
assert len(rp.statements) == 1
st = rp.statements[0]
(map2k1, mapk1) = st.agent_list()
assert map2k1.name == 'MAP2K1'
assert map2k1.db_refs['HGNC'] == '6840'
assert map2k1.db_refs['UP'] == 'Q02750'
assert mapk1.name == 'MAPK1'
assert mapk1.db_refs['HGNC'] == '6871'
assert mapk1.db_refs['UP'] == 'P28482'
assert unicode_strs(rp.statements)
def test_multiple_enzymes():
for offline in offline_modes:
rp = reach.process_text('MEK1 and MEK2 phosphorylate ERK1.',
offline=offline)
assert (len(rp.statements) == 2)
s = rp.statements[0]
if s.enz.name == 'MAP2K1':
assert (rp.statements[1].enz.name == 'MAP2K2')
else:
assert (rp.statements[1].enz.name == 'MAP2K1')
assert (s.sub.name == 'MAPK3')
s = rp.statements[1]
assert (s.sub.name == 'MAPK3')
assert unicode_strs(rp.statements)
def test_process_unicode():
for offline in offline_modes:
rp = reach.process_text('MEK1 binds ERK2\U0001F4A9.', offline=offline)
assert unicode_strs(rp.statements)
rerun = False
# Download the papers if they are not available yet
for pmcid in pmc_ids:
prefix = folder + "/" + pmcid
if not have_file(prefix + ".nxml") and not have_file(prefix + ".txt"):
txt, txt_format = get_full_text(pmcid)
if txt_format == "nxml":
fname = prefix + ".nxml"
else:
fname = prefix + ".txt"
with open(fname, "wt") as fh:
fh.write(txt.encode("utf-8"))
# Read each paper if it hasn't been read yet.
# Otherwise use the existing json extractions.
for pmcid, pmid in zip(pmc_ids, pmids):
prefix = folder + "/" + pmcid
print "Processing %s..." % pmcid
# If REACH already processed it then don't run it again
if rerun or not have_file(prefix + ".json"):
if have_file(prefix + ".txt"):
txt = open(prefix + ".txt").read().decode("utf-8")
rp = reach.process_text(txt, citation=pmid, offline=True)
elif have_file(prefix + ".nxml"):
rp = reach.process_nxml_file(prefix + ".nxml", citation=pmid, offline=True)
shutil.move("reach_output.json", prefix + ".json")
else:
rp = reach.process_json_file(prefix + ".json", citation=pmid)
run_assembly(rp.statements, folder, pmcid)
import sys
import pickle
from indra import reach
from indra.assemblers import GraphAssembler
txt = open('extension.txt', 'rt').read()
rp = reach.process_text(txt, offline=True)
st = rp.statements
for s in st:
print '%s\t%s' % (s, s.evidence[0].text)
with open('extension.pkl', 'wb') as fh:
pickle.dump(st, fh)
graphpr = {'rankdir': 'TD'}
nodepr = {'fontsize': 12, 'shape': 'plaintext', 'margin': '0,0', 'pad': 0}
ga = GraphAssembler(st, graph_properties=graphpr, node_properties=nodepr)
ga.make_model()
ga.save_dot('jnk_extension.dot')
ga.save_pdf('jnk_extension.pdf')
def test_process_unicode():
for offline in offline_modes:
rp = reach.process_text('MEK1 binds ERK2\U0001F4A9.', offline=offline)
assert unicode_strs(rp.statements)
def test_phosphorylate():
rp = reach.process_text('MEK1 phosphorylates ERK2.')
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert (s.enz.name == 'MAP2K1')
assert (s.sub.name == 'MAPK1')
def test_activity():
for offline in offline_modes:
rp = reach.process_text('MEK1 activates ERK2.', offline=offline)
assert(len(rp.statements) == 1)
assert unicode_strs(rp.statements)
def test_activate():
rp = reach.process_text('HRAS activates BRAF.')
assert(len(rp.statements) == 1)
s = rp.statements[0]
assert (s.subj.name == 'HRAS')
assert (s.obj.name == 'BRAF')
def test_bind():
rp = reach.process_text('MEK1 binds ERK2.')
assert(len(rp.statements) == 1)
def test_activity():
rp = reach.process_text('MEK1 activates ERK2.')
assert(len(rp.statements) == 1)
def test_activity():
for offline in offline_modes:
rp = reach.process_text('MEK1 activates ERK2.', offline=offline)
assert (len(rp.statements) == 1)
assert unicode_strs(rp.statements)
for pmcid in pmc_ids:
prefix = folder + '/' + pmcid
if not have_file(prefix + '.nxml') and\
not have_file(prefix + '.txt'):
txt, txt_format = get_full_text(pmcid)
if txt_format == 'nxml':
fname = prefix + '.nxml'
else:
fname = prefix + '.txt'
with open(fname, 'wt') as fh:
fh.write(txt.encode('utf-8'))
pmids.append(id_lookup(pmcid)['pmid'])
# Read each paper if it hasn't been read yet.
# Otherwise use the existing json extractions.
for pmcid, pmid in zip(pmc_ids, pmids):
prefix = folder + '/' + pmcid
print 'Processing %s...' % pmcid
# If REACH already processed it then don't run it again
if rerun or not have_file(prefix + '.json'):
if have_file(prefix + '.txt'):
txt = open(prefix + '.txt').read().decode('utf-8')
rp = reach.process_text(txt, citation=pmid)
elif have_file(prefix + '.nxml'):
rp = reach.process_nxml_file(prefix + '.nxml', citation=pmid)
shutil.move('reach_output.json', prefix + '.json')
else:
rp = reach.process_json_file(prefix + '.json', citation=pmid)
run_assembly(rp.statements, folder, pmcid)
