A machine learning-based high-precision density functional method for drug-like molecules

Models

The repo contains the models and test datasets for our aticles. The energy unit is in Hartree, The coordinate unit is in Bohr.

## DeePHF

you need first prepare the `dm_eig.npy` in data_test and do predict `l_e_delta.npy`, you can use

```

deepks test -m model.pth -o test/test -d data_test/* -D dm_eig -G

```

## DeePKS

first you should prepare the `atom.npy`, and `energy.npy` in data_test. you can test the datasets by command. 

```

deepks scf scf_input.yaml -m model.pth -s data_test -d test_out

```

# Datasets

All datasets only have `atom.npy` and `energy.npy`. The coordinate unit is `bohr`, and energy unit is `Hartree`.

## small molecules torsion

Contains 62 small molecules with 36 conformation for each under CCSD(T)/def2-TZVP.

[1] B. D. Sellers, N. C. James, A. Gobbi, A comparison of quantum and molecular mechanical methods to estimate strain energy in druglike fragments, Journal of chemical information and modeling 57 (6) (2017) 1265–127

## MPCONF91

Contains 6 molecules with 91 conformations under LNO-CCSD(T)/def2-TZVP.

[1] J. Rezac, D. Bím, O. Gutten, L. Rulisek, Toward accurate conformational energies of smaller peptides and medium-sized macrocycles: Mpconf196 benchmark energy data set, Journal of chemical theory and computation 14 (3) (2018) 1254–1

## torsionNet206

Contains 206 molecules with 4494 conformations under CCSD(T)/def2-TZVP.

[1] B. K. Rai, V. Sresht, Q. Yang, R. Unwalla, M. Tu, A. M. Mathiowetz,G. A. Bakken, Torsionnet: A deep neural network to rapidly predict small-molecule torsional energy profiles with the accuracy of quantum mechanics, Journal of Chemical Information and Modeling 62 (4) (2022) 785–80

## Out-of-plane bending

Contains 242 molecules with 3315 conformations under CCSD(T)/def2-TZVP.

[1] X. Yang, C. Liu, P. Ren, High order ab initio valence force field with chemical pattern based parameter assignment., Journal of Computational Biophysics and Chemistry 21 (4) (2021) 43

## DrugBank-T

Contains 165 molecules with 1155 conformations under CCSD(T)/def2-TZVP.

[1] V. Law, C. Knox, Y. Djoumbou, T. Jewison, A. C. Guo, Y. Liu, A. Maciejewski, D. Arndt, M. Wilson, V. Neveu, et al., Drugbank

4.0: shedding new light on drug metabolism, Nucleic acids research 42 (D1) (2014) D1091–D1097  

[2] Z. Qiao, M. Welborn, A. Anandkumar, F. R. Manby, T. F. Miller III, Orbnet: Deep learning for quantum chemistry using symmetry adapted atomic-orbital features, The Journal of chemical physics 153 (12) (2020) 124111

## Notice

if you use above datasets, please cite the original articals too