No such column pKa_relu in input file

[zengjianyou]

I successfully extracted your features, but an error occurred during my training, errors: No such column pKa_relu in input file

[zhenglz]

please show me the header of the feature file, I mean the first line of the file, The col contains the real label (experimental pKx) should be named as "pKa_true".

On Wed, Jan 20, 2021 at 3:31 PM zengjianyou [email protected] wrote:

I successfully extracted your features, but an error occurred during my training, errors: No such column pKa_relu in input file [image: image] https://user-images.githubusercontent.com/34835647/105141662-71c38c80-5b34-11eb-92e9-9855ed34b416.png

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/zhenglz/onionnet/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACN36YSX5SAX3WCKCKCWZKTS22BDLANCNFSM4WKHQWRQ .

[zengjianyou]

@zhenglz
There are still problems in extracting the data features of individual training sets， error：

Traceback (most recent call last): File "generate_features.py", line 307, in r, ele_pairs = generate_features(fn, lig_code, n_cutoffs) File "generate_features.py", line 205, in generate_features cplx.distance_pairs() File "generate_features.py", line 121, in distance_pairs self.distance_matrix_ = mt.compute_distances(self.pdb, atom_pairs=all_pairs)[0] File "/home/zjy/miniconda3/envs/drug_drug/lib/python3.6/site-packages/mdtraj/geometry/distance.py", line 62, in compute_distances pairs = ensure_type(atom_pairs, dtype=np.int32, ndim=2, name='atom_pairs', shape=(None, 2), warn_on_cast=False) File "/envs/drug_drug/lib/python3.6/site-packages/mdtraj/utils/validation.py", line 123, in ensure_type "You supplied %s" % (name, ndim, val.ndim))) ValueError: atom_pairs must be ndim 2. You supplied 1

ValueError: atom_pairs must be ndim 2. You supplied 1

[zengjianyou]

@zhenglz
Like you said, pKa_relu is added to fn_train, and pKa_True is added to test, the program can work, thank you for your help, but I found that the RMSE of the core2016 data you gave the pre-training model can only reach 1.326,WHY?

[zhenglz]

In the file "generate_features.py" line 296: n_cutoffs = np.linspace(0.1, 3.1, n_shells), please change 3.1 to 3.05. This could be error in the featurization. Please change the value, generate features, and run the prediction with v2016 coreset again, and please report the RMSE here. Thanks!

[zengjianyou]

In the file "generate_features.py" line 296: n_cutoffs = np.linspace(0.1, 3.1, n_shells), please change 3.1 to 3.05. This could be error in the featurization. Please change the value, generate features, and run the prediction with v2016 coreset again, and please report the RMSE here. Thanks!

Have you tried？ in v2016 core-set，i got 1.327 RMSE

[dagger-neo]

I successfully extracted your features, but an error occurred during my training, errors: No such column pKa_relu in input file

I am facing the same problem (no such column as pKa_relu and key error in pKa_true) even after changing the name of the columns of the input files . The command i am using to train the algorithm is as follows- python train.py -fn_train ..\onionnet-master\datasets\features_testing_v2013core_108_pka.csv -fn_validate ..\onionnet-master\datasets\features_testing_v2016core_290_pka.csv -fn_test ..\onionnet-master\datasets\input_codes_testing_core_v2016_290.csv -y_col pKa_relu pKa_true -prev_scaler models/StandardScaler.model -prev_model models/CNN_final_model_weights.h5 -log train.log -model new_tuned_model.h5 -scaler std_scaler.lib -epochs 10 -batch 32 Kindly help.