def evaluate_synthesis_trees(self, trees):
if self.celery: # Call celery worker
working = time.time()
res = evaluate_trees.apply_async(args=(trees,), kwargs={'context_scoring_method': self.context_prioritization,
'context_recommender': self.context_recommender,
'forward_scoring_method': self.forward_scoring_method,
'tree_scoring_method': self.tree_scoring_method,
'rank_threshold': self.rank_threshold_inclusion,
'prob_threshold': self.prob_threshold_inclusion,
'mincount': self.synth_mincount,
'batch_size': 500, 'n': self.max_total_contexts,
'template_count': self.template_count,
})
while not res.ready():
if int(time.time() - working) % 10 == 0:
MyLogger.print_and_log('Evaluating trees...', makeit_loc)
time.sleep(1)
evaluated_trees = res.get()
else: # Create a tree evaluation object and run it
if self.forward_scoring_method == gc.templatebased:
# nproc = number of parallel forward enumeration workers
# nproc_t = number of trees to be evaluated in parallel.
# Only use an nproc different from 1 if using the template base forward evaluation method. Otherwise
# evaluation is fast enough to do without additional
# parallelization
if len(trees) > 2 or self.parallel_tree:
nproc_t = 2
nproc = max(1, self.nproc/2)
else:
nproc_t = 1
nproc = self.nproc
else:
nproc_t = max(1, self.nproc)
nproc = 1
tree_evaluator = TreeEvaluator(
celery=False, context_recommender=self.context_recommender)
evaluated_trees = tree_evaluator.evaluate_trees(trees, context_recommender=self.context_recommender, context_scoring_method=self.context_prioritization,
forward_scoring_method=self.forward_scoring_method, tree_scoring_method=self.tree_scoring_method,
rank_threshold=self.rank_threshold_inclusion, prob_threshold=self.prob_threshold_inclusion,
mincount=self.synth_mincount, batch_size=500, n=self.max_total_contexts, nproc_t=nproc_t,
nproc=nproc, parallel=self.parallel_tree, template_count=self.template_count,
)
plausible_trees = []
print(evaluated_trees)
for tree in evaluated_trees:
if tree['plausible']:
plausible_trees.append(tree)
if plausible_trees:
MyLogger.print_and_log(
'Feasible synthesis route discovered!', makeit_loc)
else:
MyLogger.print_and_log(
'No feasible routes from buyables have been discovered. Consider changing inclusion thesholds.', makeit_loc)
return plausible_trees
def configure_coordinator(options={}, **kwargs):
if 'queues' not in options:
return
if CORRESPONDING_QUEUE not in options['queues'].split(','):
return
print('### STARTING UP A TREE EVALUATION COORDINATOR ###')
global evaluator
evaluator = TreeEvaluator(celery=True)
print('### TREE EVALUATION COORDINATOR STARTED UP ###')
print(len(truncated_mol_lib)) # for name in dataset['SMILES']:
# try:
# mol = name_parser.name_to_molecule(name)
# mol_lib.append(mol)
# except:
# continue
print("{} target molecules to expand".format(len(truncated_mol_lib)))
celery = False
NCPUS = 4
# treeBuilder = TreeBuilder(celery=celery, mincount=25, mincount_chiral=10)
Tree = MCTS(nproc=NCPUS,
mincount=gc.RETRO_TRANSFORMS_CHIRAL['mincount'],
mincount_chiral=gc.RETRO_TRANSFORMS_CHIRAL['mincount_chiral'],
celery=celery)
treeEvaluator = TreeEvaluator(context_recommender=gc.neural_network,
celery=celery)
all_routes = []
all_trees = []
all_routes_contexts = []
index_list_all_routes = []
is_first_target = True
status = 0
results_folder = "/home/hanyug/Combined_synthesis_planning/pathways/small_and_buyable/"
status_file = open(results_folder + 'status for molecules.dat', 'w')
status_file.write('SMILES\tnumberofpaths\tnotes\n')
reaction_dict = {}
for mol in truncated_mol_lib:
# for mol in [Chem.MolFromSmiles('OC(C1CCCCN1)c2cc(nc3c2cccc3C(F)(F)F)C(F)(F)F')]:
if is_first_target:
soft_reset = False