nrel / alfabet Goto Github PK

The README.md file includes the text [ml.nrel.gov](ml.nrel.gov). This is actually a broken link (missing https:// syntax), but in addition, the ml.nrel.gov site might not be the correct one. Was it intended to be [bde.ml.nrel.gov](https://bde.ml.nrel.gov/)? I.e., this website?

Hi, I was wondering if there is an option to train the model ourselves on a new set of data rather then use the trained model to make predictions?

Thanks!

absl-py==0.15.0
alfabet==0.4.1
astunparse==1.6.3
backcall==0.2.0
cached-property==1.5.2
cachetools==5.3.0
certifi==2022.12.7
charset-normalizer==3.1.0
clang==5.0
colorama==0.4.6
debugpy==1.6.6
decorator==5.1.1
entrypoints==0.4
flatbuffers==1.12
gast==0.4.0
google-auth==2.16.2
google-auth-oauthlib==0.4.6
google-pasta==0.2.0
grpcio==1.51.3
h5py==3.1.0
idna==3.4
importlib-metadata==6.1.0
ipykernel==6.16.2
ipython==7.34.0
jedi==0.18.2
joblib==1.2.0
jupyter_client==7.4.9
jupyter_core==4.12.0
keras==2.6.0
Keras-Applications==1.0.8
Keras-Preprocessing==1.1.2
libclang==15.0.6.1
Markdown==3.4.1
MarkupSafe==2.1.2
matplotlib-inline==0.1.6
nest-asyncio==1.5.6
networkx==2.6.3
nfp==0.3.12
numpy==1.19.5
oauthlib==3.2.2
opt-einsum==3.3.0
packaging==23.0
pandas==1.3.5
parso==0.8.3
pickleshare==0.7.5
Pillow==9.4.0
platformdirs==3.1.1
pooch==1.7.0
prompt-toolkit==3.0.38
protobuf==3.19.6
psutil==5.9.4
pyasn1==0.4.8
pyasn1-modules==0.2.8
Pygments==2.14.0
python-dateutil==2.8.2
pytz==2022.7.1
pywin32==305
PyYAML==6.0
pyzmq==25.0.1
rdkit==2022.9.5
requests==2.28.2
requests-oauthlib==1.3.1
rsa==4.9
scikit-learn==0.24.2
scipy==1.7.3
six==1.15.0
tensorboard==2.11.2
tensorboard-data-server==0.6.1
tensorboard-plugin-wit==1.8.1
tensorflow==2.6.0
tensorflow-estimator==2.11.0
tensorflow-io-gcs-filesystem==0.31.0
termcolor==1.1.0
tf-estimator-nightly==2.8.0.dev2021122109
threadpoolctl==3.1.0
tornado==6.2
tqdm==4.65.0
traitlets==5.9.0
typing-extensions==3.7.4.3
urllib3==1.26.15
wcwidth==0.2.6
Werkzeug==2.2.3
wincertstore==0.2
wrapt==1.12.1
zipp==3.15.0

Empty DataFrame
Columns: [bond_type, fragment1, fragment2, is_valid_stereo, bde_pred, bdfe_pred, is_valid, molecule, bond_index, bde, bdfe, set]
Index: []

I need to invoke model.predict() a lot of times. But it seems that there is a memory leak in predict().
This code gives increasing memory usage.

from alfabet import model

smiles = 'O=C(O)CCC/C=C/CC1C2CCC(O2)C1CCNC(=O)Nc1ccccc1'
while True:
	model.predict([smiles])

I have used your source code to build some BDE prediction. But it is cost too long. The main reason we believed is that you should make some additional import by local method only to boost up loading function instead of getting from scratch. Especially if it is used on a large dataset. The fastest way to boost up is that you can try to build up molecules on tops and extract them as data instead of making multiple rebuilt. Moreover, another way to boost up is that you can initialize the numpy array at large scale then fit with data instead of calling multiple np.repeat, np.stack, np.concatenate pd.stack, ....
If needs to remove duplicated you can retrieve the molecule index and used them as domain to validate radical within molecule only which is indeed so useful. Especially time-complexity is just at maximum O(N*k) where as N is the number of row and k is the number of BDE which can predict. You can built method such as predictByDefinedFile where the molecule, radicals and bond_index has been settle up to minimize calculation instead

mol_good = "Cc1ccccc1O"
model.predict([mol_good], verbose=False)
Traceback (most recent call last):
File "", line 1, in
TypeError: predict() got an unexpected keyword argument 'verbose'

Hi @pstjohn

I could not install it

I tried the following code and many other code, but it does not function

!conda create -n alfabet -c conda-forge python=3.7 rdkit
!source activate alfabet
!pip install alfabet

it gives this error

name 'alfabet' is not defined

I some other alternatives but it could not be installed

"I am working on google colab"

I hope that you could help

Cordially

The installation info could be better. I didn't get it to run at all as described using pip install, because then upon import it complains that tensorflow is missing.
If I use the setup.py instead, then the scikit module complains about min. py3.8.
Finally, after all that (and a lot of wasted time), the tests_model.py fails due to uncompiled tensorflow model:

2023-01-04 15:48:50.354177: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: SSE4.1 SSE4.2 AVX AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-01-04 15:48:52.097245: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: SSE4.1 SSE4.2 AVX AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-01-04 15:48:52.744007: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1616] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 3849 MB memory: -> device: 0, name: NVIDIA GeForce GTX 1080 Ti, pci bus id: 0000:82:00.0, compute capability: 6.1
WARNING:tensorflow:No training configuration found in save file, so the model was not compiled. Compile it manually.