czodrowskilab / machine-learning-meets-pka Goto Github PK

hi,

I tried to run the notebook, modeling.ipynb, and the previous cell units ran smoothly, but when it came to this section

est_cls = RandomForestRegressor
rf_params = dict(n_estimators=1000, n_jobs=est_jobs, verbose=verbose, random_state=seed)
name = 'RandomForest (n_estimators=1000)'

train_all_sets(est_cls, rf_params, name)

the error info:

ValueError: Input X contains NaN.
RandomForestRegressor does not accept missing values encoded as NaN natively. For supervised learning, you might want to consider sklearn.ensemble.HistGradientBoostingClassifier and Regressor which accept missing values encoded as NaNs natively. Alternatively, it is possible to preprocess the data, for instance by using an imputer transformer in a pipeline or drop samples with missing values. See https://scikit-learn.org/stable/modules/impute.html You can find a list of all estimators that handle NaN values at the following page: https://scikit-learn.org/stable/modules/impute.html#estimators-that-handle-nan-values

Here the input is made of null values, but I see that none of the previous values have null values inside them. Why is there a null value here, or am I setting the parameter wrong here?

many thanks for your help

best,

Sh-Y

MIT is a software license. For data, a data license is better (data copyright and software copyright laws are often different). May I ask you to consider making a citable Zenodo or Figshare archive of the data (novartis_cleaned_mono_unique_notraindata.sdf etc) under a CCZero license (which is quite like the MIT license but then for data)?

Hi,

I have setup the conda environment, but when I run:

python predict_sdf.py pka_ligands.sdf pka_ligands_pred.sdf
Loading SDF...
2 molecules loaded
Loading model...
Traceback (most recent call last):
  File "predict_sdf.py", line 41, in <module>
    with open('RF_CV_FMorgan3_pKa.pkl', 'rb') as f:
FileNotFoundError: [Errno 2] No such file or directory: 'RF_CV_FMorgan3_pKa.pkl'

Would you be able to share the pretrained model?

I understand if it has confidential data, this might not be possible, but I thought I would ask. :)