Skip to content

Prediction of PCE based on the embedding of pairs of molecules

Notifications You must be signed in to change notification settings

zhangxy22/Pairwise-molecular-embedding

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

81 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Pairwise molecular embedding

The following programs are designed for generating pairwise molecular embedding and modeling for power conversion efficiency(PCE) of small organic molecular solar cell.

Details of programs

FPtand is a folder containing the FPtand model that is using Text-CNN to generate molecular embedding based on donor and acceptor pair data with Mogan fingerprint in organic solar cell for predicting PCE.
The model is stored in model8_fp, and the files fa_fp.npy and fd_fp.npy are training data of acceptor and donor, respectively.
fp_Y.npy is the file containing target PCE values, and fp1.py is the main program code.

FPpara is a folder containing a similar model as FPtand except that manipulates the donor and acceptor data in parallel to make a final prediction.

MLemb includes four baseline manchine learning models: RF, SVR, Xgboost and gcforest. The inputs for the training and test of models are featuremap_train.csv and featuremap_test.csv, respectively.

fingerprint includes FPtand models with embedding from four types of fingerprints, namely APfp, CDKfp, GRAPHfp and MACCS.

visualization consists of t-SNE and Shap analyses for embedding.

Environment

Configuring the environment according to FPtand.yml before running the code.

About

Prediction of PCE based on the embedding of pairs of molecules

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages