scores = defaultdict(list)
# loop over repetitions
for q in range(conf.num_reps):
training_list = cPickle.load(training_input)
test_list = [i for i in range(num_actives) if i not in training_list[:num_query_mols]]
test_list += [i for i in range(num_decoys) if i not in training_list[num_query_mols:]]
# loop over fps
single_score = defaultdict(list)
for fp in fp_names:
query_fps = [actives[i][1][fp] for i in training_list[:num_query_mols]]
# test_list: first actives then decoys
test_fps = [[actives[i][0], actives[i][1][fp], 1] for i in test_list[:num_test_actives]]
test_fps += [[decoys[i][0], decoys[i][1][fp], 0] for i in test_list[num_test_actives:]]
for tmp_mol in test_fps:
tmp_score = scor.getBulkSimilarity(tmp_mol[1], query_fps, simil_metric)
# use max fusion
# store : [similarity, internal ID, active/inactive]
single_score[fp].append([tmp_score[0], tmp_mol[0], tmp_mol[2]])
# rank list according to similarity
scores[fp].append(sorted(single_score[fp], reverse=True))
# write scores to file
if do_append:
outfile = gzip.open(outpath+'/list_'+dataset+'_'+str(target)+'.pkl.gz', 'ab+') # binary format
else:
outfile = gzip.open(outpath+'/list_'+dataset+'_'+str(target)+'.pkl.gz', 'wb+') # binary format
for fp in fp_names:
cPickle.dump([fp, scores[fp]], outfile, 2)
outfile.close()
print "scoring done and scored lists written"
]
np_test_fps = [
np_fps_act[i] for i in test_list[:num_test_actives]
]
np_test_fps += [
np_fps_dcy[i] for i in test_list[num_test_actives:]
]
test_mols = [[actives[i][0], 1]
for i in test_list[:num_test_actives]]
test_mols += [[decoys[i][0], 0]
for i in test_list[num_test_actives:]]
# calculate similarity with standard fp
std_simil = []
for fp in test_fps:
tmp_simil = scor.getBulkSimilarity(fp, train_fps,
simil_metric)
tmp_simil.sort(reverse=True)
std_simil.append(tmp_simil[0])
# rank based on probability (and second based on similarity)
single_score = ml.predict_proba(np_test_fps)
# store: [probability, similarity, internal ID, active/inactive]
single_score = [[
m[1], s, t[0], t[1]
] for m, s, t in zip(single_score, std_simil, test_mols)]
single_score.sort(reverse=True)
scores['rf_' + fp_build].append(single_score)
# write scores to file
if do_append:
outfile = gzip.open(outpath + '/list_' + dataset + '_' +
if i not in training_list[num_query_mols:]
]
# loop over fps
single_score = defaultdict(list)
for fp in fp_names:
query_fps = [
actives[i][1][fp]
for i in training_list[:num_query_mols]
]
# test_list: first actives then decoys
test_fps = [[actives[i][0], actives[i][1][fp], 1]
for i in test_list[:num_test_actives]]
test_fps += [[decoys[i][0], decoys[i][1][fp], 0]
for i in test_list[num_test_actives:]]
for tmp_mol in test_fps:
tmp_score = scor.getBulkSimilarity(
tmp_mol[1], query_fps, simil_metric)
# use max fusion
# store : [similarity, internal ID, active/inactive]
single_score[fp].append(
[tmp_score[0], tmp_mol[0], tmp_mol[2]])
# rank list according to similarity
scores[fp].append(sorted(single_score[fp], reverse=True))
# write scores to file
if do_append:
outfile = gzip.open(outpath + '/list_' + dataset + '_' +
str(target) + '.pkl.gz',
'ab+') # binary format
else:
outfile = gzip.open(outpath + '/list_' + dataset + '_' +
str(target) + '.pkl.gz',
np_train_fps += [np_fps_dcy[i] for i in training_list[num_query_mols:]]
# fit random forest
ml.fit(np_train_fps, ys_fit)
# test fps and molecule info
test_fps = [actives[i][1] for i in test_list[:num_test_actives]]
test_fps += [decoys[i][1] for i in test_list[num_test_actives:]]
np_test_fps = [np_fps_act[i] for i in test_list[:num_test_actives]]
np_test_fps += [np_fps_dcy[i] for i in test_list[num_test_actives:]]
test_mols = [[actives[i][0], 1] for i in test_list[:num_test_actives]]
test_mols += [[decoys[i][0], 0] for i in test_list[num_test_actives:]]
# calculate similarity with standard fp
std_simil = []
for fp in test_fps:
tmp_simil = scor.getBulkSimilarity(fp, train_fps, simil_metric)
tmp_simil.sort(reverse=True)
std_simil.append(tmp_simil[0])
# rank based on probability (and second based on similarity)
single_score = ml.predict_proba(np_test_fps)
# store: [probability, similarity, internal ID, active/inactive]
single_score = [[m[1], s, t[0], t[1]] for m,s,t in zip(single_score,std_simil,test_mols)]
single_score.sort(reverse=True)
scores['rf_'+fp_build].append(single_score)
# write scores to file
if do_append:
outfile = gzip.open(outpath+'/list_'+dataset+'_'+str(target)+'.pkl.gz', 'ab+') # binary format
else:
outfile = gzip.open(outpath+'/list_'+dataset+'_'+str(target)+'.pkl.gz', 'wb+') # binary format