A DEEP LEARNING BASED METHOD FOR ESTIMATING EFFECTIVE PROTEIN-LIGAND AFFINITY

prerequirment:

anaconda python 2.7 tensorflow keras scklearn numpy pandas

################## ################# For testing cases that have experimental value step 1: unzip the all_data/data.zip, by cd all_data; unzip data.zip

step 2: got the performance estimator(r value, rmse, etc) by: python deep_learn_rob_residual_zhpxxx_n_regression_load_drop50.py

step 3: list the vina score, the DeepBindRG prediction value, and the experimental value (all_energies.sort is the output file from vina docking) python perform_vina.py

check the output file out_list.csv

collumn 1-4 are : name,experiment value, DeepBindRG value, Vina value

############################ ############################ For application cases that have no experimental value:

python deep_learn_rob_residual_zhpxxx_n_regression_load_drop50_use.py

check the out_file.csv

Citation： DeepBindRG: a deep learning based method for estimating effective protein-ligand affinity

If there is any technique problem, please no hestitated to contact by email hp.zhang@siat.ac.cn.