|
1 | 1 | Uni-Mol: A Universal 3D Molecular Representation Learning Framework
|
2 | 2 | ===================================================================
|
3 | 3 |
|
4 |
| -[[Paper](https://openreview.net/forum?id=6K2RM6wVqKu)], [[Uni-Mol Docking Colab](https://colab.research.google.com/github/dptech-corp/Uni-Mol/blob/main/unimol/notebooks/unimol_binding_pose_demo.ipynb)] |
| 4 | +[[Paper](https://openreview.net/forum?id=6K2RM6wVqKu)], [[Uni-Mol Docking Colab](https://colab.research.google.com/github/deepmodeling/Uni-Mol/blob/main/unimol/notebooks/unimol_binding_pose_demo.ipynb)] |
5 | 5 |
|
6 | 6 | Authors: Gengmo Zhou, Zhifeng Gao, Qiankun Ding, Hang Zheng, Hongteng Xu, Zhewei Wei, Linfeng Zhang, Guolin Ke
|
7 | 7 |
|
@@ -63,23 +63,23 @@ Uni-Mol's pretrained model weights
|
63 | 63 |
|
64 | 64 | | Model | File Size |Update Date | Download Link |
|
65 | 65 | |--------------------------|------------| ------------|--------------------------------------------------------------|
|
66 |
| -| molecular pretrain | 181MB | Aug 17 2022 |https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/mol_pre_no_h_220816.pt | |
67 |
| -| pocket pretrain | 181MB | Aug 17 2022 |https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/pocket_pre_220816.pt | |
| 66 | +| molecular pretrain | 181MB | Aug 17 2022 |https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/mol_pre_no_h_220816.pt | |
| 67 | +| pocket pretrain | 181MB | Aug 17 2022 |https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/pocket_pre_220816.pt | |
68 | 68 |
|
69 | 69 |
|
70 | 70 | Uni-Mol's finetuned model weights
|
71 | 71 | ----------------------------------
|
72 | 72 |
|
73 | 73 | | Model | File Size| Update Date| Download Link |
|
74 | 74 | |-------------------------------------------------|---------| -----------|--------------------------------------------------------------------|
|
75 |
| -| molecular conformation generation (qm9) | 181MB | Sep 8 2022 |https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/qm9_220908.pt | |
76 |
| -| molecular conformation generation (drugs) | 181MB | Sep 8 2022 |https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/drugs_220908.pt | |
77 |
| -| Protein-ligand binding pose prediction | 415MB | Sep 8 2022 |https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/binding_pose_220908.pt | |
| 75 | +| molecular conformation generation (qm9) | 181MB | Sep 8 2022 |https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/qm9_220908.pt | |
| 76 | +| molecular conformation generation (drugs) | 181MB | Sep 8 2022 |https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/drugs_220908.pt | |
| 77 | +| Protein-ligand binding pose prediction | 415MB | Sep 8 2022 |https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/binding_pose_220908.pt | |
78 | 78 |
|
79 | 79 |
|
80 | 80 | Dependencies
|
81 | 81 | ------------
|
82 |
| - - [Uni-Core](https://github.com/dptech-corp/Uni-Core), check its [Installation Documentation](https://github.com/dptech-corp/Uni-Core#installation). |
| 82 | + - [Uni-Core](https://github.com/deepmodeling/Uni-Core), check its [Installation Documentation](https://github.com/deepmodeling/Uni-Core#installation). |
83 | 83 | - rdkit==2022.9.3, install via `pip install rdkit-pypi==2022.9.3`
|
84 | 84 |
|
85 | 85 | To use GPUs within docker you need to [install nvidia-docker-2](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker) first. Use the following command to pull the docker image:
|
@@ -268,7 +268,7 @@ For ClinTox, Tox21, ToxCast, SIDER, HIV, PCBA and MUV, we set `loss_func=multi_t
|
268 | 268 | For ESOL, FreeSolv and Lipo, we set `loss_func=finetune_mse`.
|
269 | 269 | For QM7, QM8 and QM9, we set `loss_func=finetune_smooth_mae`.
|
270 | 270 |
|
271 |
| -**NOTE**: Our first version of the molecular pretraining ran with **all hydrogen** pretrained model, and above hyper-parameters are also for **all hydrogen** pretrained model. You can download the [all hydrogen model parameter](https://github.com/dptech-corp/Uni-Mol/releases/download/v0.1/mol_pre_all_h_220816.pt) here, and use it with `only_polar=-1` to reproduce our results. The performance of pretraining model with **no hydrogen** is very close to the **all hydrogen** one in molecular property prediction. We will update the hyperparameters for the no hydrogen version later. |
| 271 | +**NOTE**: Our first version of the molecular pretraining ran with **all hydrogen** pretrained model, and above hyper-parameters are also for **all hydrogen** pretrained model. You can download the [all hydrogen model parameter](https://github.com/deepmodeling/Uni-Mol/releases/download/v0.1/mol_pre_all_h_220816.pt) here, and use it with `only_polar=-1` to reproduce our results. The performance of pretraining model with **no hydrogen** is very close to the **all hydrogen** one in molecular property prediction. We will update the hyperparameters for the no hydrogen version later. |
272 | 272 |
|
273 | 273 | **NOTE**: For reproduce, you can do the validation on test set while training, with `--valid-subset valid` changing to `--valid-subset valid,test`. The model selection is still based on the performance of the valid set. It is controlled by `--best-checkpoint-metric $metric`.
|
274 | 274 |
|
@@ -537,4 +537,4 @@ Please kindly cite this paper if you use the data/code/model.
|
537 | 537 | License
|
538 | 538 | -------
|
539 | 539 |
|
540 |
| -This project is licensed under the terms of the MIT license. See [LICENSE](https://github.com/dptech-corp/Uni-Mol/blob/main/LICENSE) for additional details. |
| 540 | +This project is licensed under the terms of the MIT license. See [LICENSE](https://github.com/deepmodeling/Uni-Mol/blob/main/LICENSE) for additional details. |
0 commit comments