Descriptor NaN

[mikewzp]

Hi, I'm getting trouble with computing descriptors.

[MaSIF-search]

./compute_descriptors.sh 1AKJ_AB_DE
Traceback (most recent call last):
File "/media/ps/3632610B3260D203/DeepLearning/masif/source/masif_ppi_search/masif_ppi_search_comp_desc.py", line 233, in
roc_auc = 1.0 - compute_roc_auc(pos_dists, neg_dists)
File "/media/ps/3632610B3260D203/DeepLearning/masif/source/masif_ppi_search/masif_ppi_search_comp_desc.py", line 21, in compute_roc_auc
return metrics.roc_auc_score(labels, dist_pairs)
File "/home/ps/.conda/envs/masif/lib/python3.6/site-packages/sklearn/utils/validation.py", line 63, in inner_f
return f(*args, **kwargs)
File "/home/ps/.conda/envs/masif/lib/python3.6/site-packages/sklearn/metrics/_ranking.py", line 524, in roc_auc_score
y_score = check_array(y_score, ensure_2d=False)
File "/home/ps/.conda/envs/masif/lib/python3.6/site-packages/sklearn/utils/validation.py", line 63, in inner_f
return f(*args, **kwargs)
File "/home/ps/.conda/envs/masif/lib/python3.6/site-packages/sklearn/utils/validation.py", line 721, in check_array
allow_nan=force_all_finite == 'allow-nan')
File "/home/ps/.conda/envs/masif/lib/python3.6/site-packages/sklearn/utils/validation.py", line 106, in _assert_all_finite
msg_dtype if msg_dtype is not None else X.dtype)
ValueError: Input contains NaN, infinity or a value too large for dtype('float32').

According to the traceback, find that: during compute_roc_auc(pos_dists, neg_dists) these two descriptor files exist but values are NAN.
# Save descriptors
np.save(os.path.join(out_desc_dir, "p1_desc_straight.npy"), desc1_str)
np.save(os.path.join(out_desc_dir, "p1_desc_flipped.npy"), desc1_flip)
# Compute pos_dists
pos_dists = np.sqrt(np.sum(np.square(desc1_str[k1] - desc2_flip[k2]), axis=1))
neg_dists = np.sqrt( np.sum(np.square(desc1_str[kneg1] - desc2_flip[kneg2]), axis=1) )

[MaSIF-Site]

• predict_site.sh
• color_site.sh
  The output of the predict_site script is a contact score calculated by the Neural Network between 0 and 1 for each center point. However, these files are also exist but NAN values.
  (masif) ps@ps:~/DeepLearning/masif/data/masif_site$ ls output/all_feat_3l/pred_data/
  pred_1A0G_A.npy pred_1A0G_B.npy pred_1A0H_D.npy pred_1A0H_E.npy

Does anyone experience similar problems? What the standard output should be?
Any comment or ideas would be appreciated!