def map_metabolites_compartments(model_id2dfs, chebi=None):
if not chebi:
chebi = parse_simple(get_chebi())
model_id2m_ids_groups = map_metabolites(model_id2dfs, chebi)
model_id2c_ids_groups = map_comps(model_id2dfs)
model_id2c_id2i = defaultdict(dict)
for i, model_id2c_ids in enumerate(model_id2c_ids_groups):
for model_id, c_ids in model_id2c_ids.items():
model_id2c_id2i[model_id].update({c_id: i for c_id in c_ids})
model_id2m_ids_same_comp_groups = []
for model_id2m_ids in model_id2m_ids_groups:
i2m_ids = defaultdict(list)
for model_id, m_ids in model_id2m_ids.items():
for m_id in m_ids:
df, _, _ = model_id2dfs[model_id]
c_id = df.at[m_id, 'Compartment']
if c_id in model_id2c_id2i[model_id]:
i2m_ids[model_id2c_id2i[model_id][c_id]].append((model_id, m_id))
for m_ids in i2m_ids.values():
model_id2m_ids = defaultdict(set)
if len(m_ids) > 1:
for model_id, m_id in m_ids:
model_id2m_ids[model_id].add(m_id)
if model_id2m_ids:
model_id2m_ids_same_comp_groups.append(model_id2m_ids)
return model_id2c_ids_groups, [it for it in model_id2m_ids_same_comp_groups if len(it) > 1], model_id2c_id2i
def import_sbml(input_model, sbml_file):
logging.info('parsing ChEBI')
chebi = parse_simple(get_chebi())
logging.info('reading generalized model from %s' % sbml_file)
r_id2g_id, s_id2gr_id, ub_sps = get_quotient_maps(chebi, sbml_file)
logging.info('fixing labels and compartments')
check_names(input_model)
check_compartments(input_model)
logging.info('annotating with GO')
go = parse_simple(get_go())
annotate_compartments(input_model, go)
c_id2level = comp2level(input_model, go)
c_id2info, c_id2outs = compute_c_id2info(c_id2level, input_model)
def get_r_comp(all_comp_ids):
if len(all_comp_ids) == 1:
return all_comp_ids.pop()
get_level = lambda c_id: c_id2level[c_id][0]
outer_most = min(all_comp_ids, key=get_level)
inner_most = max(all_comp_ids, key=get_level)
outer_level, inner_level = get_level(outer_most), get_level(inner_most)
if outer_level == inner_level or (outer_most
not in c_id2outs[inner_most]):
candidates = set(c_id2outs[inner_most]) & set(
c_id2outs[outer_most])
if candidates:
return max(candidates, key=get_level)
else:
return outer_most
if inner_level - outer_level > 1:
return max(c_id2outs[inner_most], key=get_level)
return outer_most
logging.info('initialising the graph')
graph = tlp.newGraph()
graph.setName(input_model.getId())
create_props(graph)
logging.info('adding species nodes')
id2n = species2nodes(graph, input_model, ub_sps)
logging.info('adding reaction nodes')
reactions2nodes(get_r_comp, graph, id2n, input_model)
# for n in (n for n in graph.getNodes() if TYPE_SPECIES == graph[TYPE][n] and graph.deg(n) > 5 \
# and not graph[ID][n] in s_id2gr_id):
# graph[UBIQUITOUS][n] = True
logging.info('duplicating nodes')
duplicate_nodes(graph)
logging.info('marking species/reaction groups')
mark_ancestors(graph, r_id2g_id, s_id2gr_id, c_id2info)
return graph, c_id2info, c_id2outs, chebi, ub_sps
def add_boundary_metabolites(in_sbml, out_sbml):
"""
Creates a boundary compartment with the id 'Boundary',
and transforms each input/output reaction '*_e <-> ' into '*_e <-> *_b'.
:param in_sbml: path to the SBML file with the original model
:param out_sbml: path where to store the resulting SBML file
"""
input_doc = libsbml.SBMLReader().readSBML(in_sbml)
model = input_doc.getModel()
chebi = parse_simple(get_chebi())
annotate_metabolites(model, chebi)
create_boundary_species_in_boundary_reactions(model)
libsbml.SBMLWriter().writeSBMLToFile(input_doc, out_sbml)
def combine_models(model_id2sbml, model_id2S, path):
logging.info('Going to merge models...')
chebi = parse_simple(get_chebi())
model_id2dfs = get_model_data(model_id2sbml)
model_id2c_id_groups, model_id2m_id_groups, model_id2c_id2i = \
map_metabolites_compartments(model_id2dfs, chebi=chebi)
logging.info('Mapped metabolites and compartments.')
ignore_m_ids = get_ignored_metabolites(model_id2dfs, get_proton_ch_ids())
S = join(model_id2m_id_groups, model_id2S)
ignore_m_ids |= {S.m_id2gr_id[m_id] for m_id in ignore_m_ids if m_id in S.m_id2gr_id}
merge(S, ignore_m_ids)
model_id2r_id_groups = get_r_id_groups(S)
logging.info('Mapped reactions.')
sbml = os.path.join(path, 'Merged_model.xml')
model_id2id2id, common_ids, S = simple_merge_models(S, model_id2c_id2i, model_id2dfs, sbml)
return sbml, S, model_id2id2id, common_ids, model_id2dfs, \
(model_id2c_id_groups, model_id2m_id_groups, model_id2r_id_groups)
def merge_models(in_sbml_list, out_sbml):
if not in_sbml_list:
raise ValueError('Provide SBML models to be merged')
go = parse_simple(get_go())
chebi = parse_simple(get_chebi())
i = 0
model_ids = set()
go2c_id = {}
doc = libsbml.SBMLDocument(2, 4)
model = doc.createModel()
model.setId('m_merged')
m_c_ids = set()
for o_sbml in in_sbml_list:
o_doc = libsbml.SBMLReader().readSBML(o_sbml)
set_consistency_level(o_doc)
o_doc.checkL2v4Compatibility()
o_doc.setLevelAndVersion(2, 4, False, True)
o_model = o_doc.getModel()
logging.info("Processing %s" % o_sbml)
model_id = get_model_id(i, model_ids, o_model)
update_model_element_ids(model_id, o_model, go2c_id, go, chebi)
for e in o_model.getListOfCompartments():
c_id = e.getId()
if c_id not in m_c_ids:
if model.addCompartment(e):
copy_compartment(e, model)
m_c_ids.add(c_id)
for e in o_model.getListOfSpecies():
if model.getSpecies(e.getId()):
continue
if model.addSpecies(e):
copy_species(e, model)
for e in o_model.getListOfReactions():
if model.addReaction(e):
copy_reaction(e, model)
libsbml.writeSBMLToFile(doc, out_sbml)
<body>
<br/>
<p class="centre">We are visualizing your model now...</p>
<br/>
<img class="img-centre" src="http://mimoza.bordeaux.inria.fr/lib/modelmap/ajax-loader.gif" id="img" />
<div id="hidden" style="visibility:hidden;height:0px;">''')
sys.stdout.flush()
temp = os.dup(sys.stdout.fileno())
try:
url = '/%s/comp.html' % m_dir_id
if not os.path.exists(os.path.join('..', m_dir_id, 'comp.html')):
chebi = parse_simple(get_chebi())
reader = libsbml.SBMLReader()
input_document = reader.readSBML(groups_sbml)
input_model = input_document.getModel()
# sbml -> tulip graph
logging.info('sbml -> tlp')
graph, c_id2info, c_id2outs, chebi, ub_sps = import_sbml(
input_model, groups_sbml)
try:
n2xy = parse_layout_sbml(groups_sbml)
except LoPlError:
n2xy = None
fc, (n2lo,
<body>
<br/>
<p class="centre">We are visualizing your model now...</p>
<br/>
<img class="img-centre" src="http://mimoza.bordeaux.inria.fr/lib/modelmap/ajax-loader.gif" id="img" />
<div id="hidden" style="visibility:hidden;height:0px;">''')
sys.stdout.flush()
temp = os.dup(sys.stdout.fileno())
try:
url = '/%s/comp.html' % m_dir_id
if not os.path.exists(os.path.join('..', m_dir_id, 'comp.html')):
chebi = parse_simple(get_chebi())
reader = libsbml.SBMLReader()
input_document = reader.readSBML(groups_sbml)
input_model = input_document.getModel()
# sbml -> tulip graph
logging.info('sbml -> tlp')
graph, c_id2info, c_id2outs, chebi, ub_sps = import_sbml(input_model, groups_sbml)
try:
n2xy = parse_layout_sbml(groups_sbml)
except LoPlError:
n2xy = None
fc, (n2lo, e2lo), hidden_c_ids, c_id_hidden_ubs = graph2geojson(c_id2info, c_id2outs, graph, n2xy, onto=chebi)
c_id2out_c_id = {}
__author__ = 'anna'
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser(description="Generalizes an SBML model.")
parser.add_argument('--model', required=True, type=str,
help="input model in SBML format")
parser.add_argument('--output_model', default=None, type=str,
help="path to the output generalized model in SBML format")
parser.add_argument('--groups_model', default=None, type=str,
help="path to the output model in SBML format with groups extension to encode similar elements")
parser.add_argument('--verbose', action="store_true", help="print logging information")
parser.add_argument('--log', default=None, help="a log file")
params = parser.parse_args()
prefix = os.path.splitext(params.model)[0]
if not params.output_model:
params.output_model = "%s_generalized.xml" % prefix
if not params.groups_model:
params.groups_model = "%s_with_groups.xml" % prefix
if params.verbose:
logging.basicConfig(level=logging.INFO)
logging.info("parsing ChEBI...")
ontology = parse_simple(get_chebi())
r_id2clu, s_id2clu, _, _ = generalize_model(params.model, ontology, params.groups_model, params.output_model,
ub_chebi_ids={'chebi:ch'})
def process_sbml(sbml,
verbose,
ub_ch_ids=None,
web_page_prefix=None,
generalize=True,
log_file=None,
id2mask=None,
layer2mask=DEFAULT_LAYER2MASK,
tab2html=None,
title=None,
h1=None,
id2color=None,
tabs={ABOUT_TAB, DOWNLOAD_TAB},
info='',
invisible_layers=None,
sbgn=True,
cytoscape=True):
"""
Generalizes and visualizes a given SBML model.
:param sbml: a path to the input SBML file
:param verbose: if logging information should be printed
:param ub_ch_ids: optional, ChEBI ids to be considered as ubiquitous. If left None, will be calculated automatically.
:param web_page_prefix: optional, how this model's webpage will be identified.
If left None an identifier will be generated based on the SBML file's md5.
:param generalize: optional, whether the generalization should be performed. The default is True
:param log_file: optional, a file where the logging information should be redirected
(only needed if verbose is set to True)
:param id2mask: optional,
:param layer2mask: optional, a dict storing the correspondence between a layer name and an its id mask
:param tab2html: optional,
:param title: optional, the title for the web page
:param h1: optional, the main header of the web page
:param id2color: optional,
:param tabs: optional, a set of names of tabs that should be shown
:param info: optional, additional information to be displayed in the bottom of the web page
:param invisible_layers: optional, the layers of the visualized metabolic map that should be hidden
:return: void
"""
# Read the SBML
reader = libsbml.SBMLReader()
doc = reader.readSBML(sbml)
model = doc.getModel()
if not model:
raise Exception(
"The model should be in SBML format, check your file %s" % sbml)
model_id = model.getId()
if not model_id:
sbml_name = os.path.splitext(os.path.basename(sbml))[0]
model.setId(sbml_name)
model_id = sbml_name
# Prepare the output directories
web_page_prefix = web_page_prefix if web_page_prefix else check_md5(sbml)
sbml_dir = dirname(abspath(sbml))
directory = os.path.join(sbml_dir, web_page_prefix)
if not os.path.exists(directory):
os.makedirs(directory)
lib_path = os.path.join(directory, 'lib')
if not os.path.exists(lib_path):
copytree(get_lib(), lib_path)
# Prepare the logger
if verbose:
logging.captureWarnings(True)
logging.basicConfig(level=logging.INFO,
format='%(asctime)s: %(message)s',
datefmt="%Y-%m-%d %H:%M:%S",
filename=log_file)
# Generalize the model if needed
groups_sbml = os.path.join(directory, '%s_with_groups.xml' % model_id)
gen_sbml = os.path.join(directory, '%s_generalized.xml' % model_id)
if check_for_groups(sbml, SBO_CHEMICAL_MACROMOLECULE,
GROUP_TYPE_UBIQUITOUS):
if sbml != groups_sbml:
if not libsbml.SBMLWriter().writeSBMLToFile(doc, groups_sbml):
raise Exception("Could not write your model to %s" %
groups_sbml)
else:
chebi = parse_simple(get_chebi())
if generalize:
logging.info('Generalizing the model...')
generalize_model(sbml,
chebi,
groups_sbml,
gen_sbml,
ub_chebi_ids=ub_ch_ids)
else:
gen_sbml = None
logging.info('Ubiquitizing the model...')
ubiquitize_model(sbml, chebi, groups_sbml, ub_chebi_ids=ub_ch_ids)
# Visualize the model
reader = libsbml.SBMLReader()
input_document = reader.readSBML(groups_sbml)
input_model = input_document.getModel()
root, c_id2info, c_id2outs, chebi, ub_sps = import_sbml(
input_model, groups_sbml)
c_id2out_c_id = {}
for c_id, c_info in c_id2info.items():
_, _, (_, out_c_id) = c_info
if out_c_id:
c_id2out_c_id[c_id] = out_c_id
try:
n2xy = parse_layout_sbml(sbml)
if n2xy:
logging.info('Found layout in the model...')
r_size = next((n2xy[r.getId()][1][0]
for r in input_model.getListOfReactions()
if r.getId() in n2xy), None)
if r_size:
scale_factor = REACTION_SIZE / r_size
if scale_factor != 1:
keys = n2xy.keys()
for n_id in keys:
value = n2xy[n_id]
if isinstance(value, dict):
value = {
r_id:
(scale(xy,
scale_factor), scale(wh, scale_factor))
for (r_id, (xy, wh)) in value.items()
}
else:
xy, wh = value
value = scale(xy, scale_factor), scale(
wh, scale_factor)
n2xy[n_id] = value
except LoPlError:
n2xy = None
fc, (n2lo, e2lo), hidden_c_ids, c_id_hidden_ubs = \
graph2geojson(c_id2info, c_id2outs, root, n2xy, id2mask=id2mask, onto=chebi,
colorer=color if not id2color else lambda graph: color_by_id(graph, id2color))
if n2lo:
groups_document = reader.readSBML(groups_sbml)
groups_model = groups_document.getModel()
if gen_sbml:
gen_document = reader.readSBML(gen_sbml)
gen_model = gen_document.getModel()
else:
gen_model = False
save_as_layout_sbml(groups_model, gen_model, groups_sbml, gen_sbml,
n2lo, ub_sps)
if sbgn:
groups_sbgn = os.path.join(directory, '%s.sbgn' % model_id)
gen_sbgn = os.path.join(directory,
'%s_generalized.sbgn' % model_id)
try:
save_as_sbgn(n2lo, e2lo, groups_model, groups_sbgn)
logging.info(' exported as SBGN %s' % groups_sbgn)
except Exception as e:
logging.error("Didn't manage to save to SBGN: %s" % e)
if gen_model:
try:
save_as_sbgn(n2lo, e2lo, gen_model, gen_sbgn)
logging.info(' exported as SBGN %s' % groups_sbgn)
except Exception as e:
logging.error("Didn't manage to save to SBGN: %s" % e)
if cytoscape:
out_json = os.path.join(directory, '%s.cyjs' % model_id)
save_as_cytoscape_json(n2lo, model, out_json, ub_sps)
logging.info(' exported as Cytoscape json %s' % out_json)
if gen_model:
out_json = os.path.join(directory,
'%s_generalized.cyjs' % model_id)
save_as_cytoscape_json(n2lo, gen_model, out_json, ub_sps)
# Serialize the result
serialize(directory=directory,
m_dir_id=web_page_prefix,
input_model=input_model,
c_id2level2features=fc,
c_id2out_c_id=c_id2out_c_id,
hidden_c_ids=hidden_c_ids,
c_id_hidden_ubs=c_id_hidden_ubs,
tabs=tabs,
groups_sbml=groups_sbml,
layer2mask=layer2mask,
tab2html=tab2html,
title=title,
h1=h1,
invisible_layers=invisible_layers)
parser.add_argument('--model', required=True, type=str,
help="input model in SBML format "
"or a directory containing several models if they first need to be merged")
parser.add_argument('--chebi', default=None, type=str, help="path to the ChEBI ontology file in OBO format")
parser.add_argument('--output_model', default=None, type=str,
help="path to the output generalized model in SBML format")
parser.add_argument('--groups_model', default=None, type=str,
help="path to the output model in SBML format with groups extension to encode similar elements")
parser.add_argument('--merged_model', default=None, type=str,
help="path to the output merged model in SBML format")
parser.add_argument('--verbose', action="store_true", help="print logging information")
parser.add_argument('--log', default=None, help="a log file")
params = parser.parse_args()
if not params.chebi:
params.chebi = get_chebi()
prefix = os.path.splitext(params.model)[0]
if not params.merged_model:
params.merged_model = "%s.xml" % prefix
if not params.output_model:
params.output_model = "%s_generalized.xml" % prefix
if not params.groups_model:
params.groups_model = "%s_with_groups.xml" % prefix
if params.verbose:
logging.basicConfig(level=logging.INFO)
logging.info("parsing ChEBI...")
ontology = parse_simple(params.chebi)
if os.path.isdir(params.model):
in_sbml_list = ['%s/%s' % (params.model, f) for f in glob.glob(os.path.join(params.model, '*'))
