Skip to content
/ CMGN Public

A Conditional Molecular Generation Net to Design Target-specific Molecules with Desired Properties

Notifications You must be signed in to change notification settings

WJmodels/CMGN

Repository files navigation

CMGN

A Conditional Molecular Generation Net to Design Target-specific Molecules with Desired Properties

environment

rdkit

torchtext

torch

torchvision

transformers

datasets

tqdm

wandb

pytorch_lightning

download

Model weights and datasets:

https://pan.baidu.com/s/19I4_gQyWd0x1CBkNh55YiA nfcp

test of molecular generation for specific-target

python test_infer_molecular_formula_mul_pr_rerank_copy_aidd_zinc1.py \
--val_folder = "exp/test"
--weights_path = "weight_smiles_decoder/PKI_100epoch_BTK_200epoch/epoch_199_loss_0.024621.pth"
--use_dict = {"fragment": True, "molecular_weight": True, "SMILES": True,
                "logP": False, "QED": True, "SA": True}
--save_path = "exp_result/BTK_2_3_PKI100btk200_allprop.cmc"

About

A Conditional Molecular Generation Net to Design Target-specific Molecules with Desired Properties

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages