search

e-yi / pafnucy_pytorch

a pytorch implementation of Pafnucy.
8 stars 0 forks source link

How to calculate the scaler for a new protein target dataset? #1

Open Dadiao-shuai opened 11 months ago

Dadiao-shuai commented 11 months ago

I am using a new protein dataset and using pafnucy predict the score. As the Pafnucy README said, I need to specify the --charge_scaler if I am using a dataset different from PDBbindv2016. BUT how can I calculating my new dataset?

SanFran-Me commented 11 months ago

I want ro predict on PDBbind v2020. However, I do not know how to calculate their charge-scaler to be set when predicting.

Kerro-junior commented 11 months ago

my python script to calculate the partial charge distribution of ligands is :

ligand_files = [os.path.join(cplx_dir, f) for f in os.listdir(cplx_dir) if f.endswith('.mol2')]
all_charges_array = []
for file_path in ligand_files:
    partial_charges = extract_partial_charges(file_path)
    all_charges_array.extend(partial_charges)
all_charges_np_array = np.array(all_charges_array)
# Calculate the standard deviation of the distribution
sd = np.std(all_charges_np_array)

but the result of PDBbindv2016_general+refined is about 0.2, not 0.4 as the Pafnucy default setting. @e-yi