Log In Sign Up
Models:
Amanda Davis
Amanda-D/
DrugCLIP
0
Downloads: 1
Diff of /README.md [b40915] .. [8956d4]

Switch to unified view

a/README.md b/README.md
1 
# DrugCLIP: Contrastive Protein-Molecule Representation Learning for Virtual Screening
 1 
# DrugCLIP: Contrastive Protein-Molecule Representation Learning for Virtual Screening
2 

  2
  





  3
  
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://github.com/xxxx/blob/main/LICENSE)

  3
  
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://github.com/xxxx/blob/main/LICENSE)






  4
  
[![ArXiv](http://img.shields.io/badge/cs.LG-arXiv%3A2310.06367-B31B1B.svg)](https://arxiv.org/pdf/2310.06367.pdf)

  4
  
[![ArXiv](http://img.shields.io/badge/cs.LG-arXiv%3A2310.06367-B31B1B.svg)](https://arxiv.org/pdf/2310.06367.pdf)





  5
  

  5
  





  6
  
<!-- [[Code](xxxx - Overview)] -->

  6
  
<!-- [[Code](xxxx - Overview)] -->





  7
  

  7
  





  8
  
![cover](framework.png)

  8
  
![cover](https://github.com/bowen-gao/DrugCLIP/blob/main/framework.png?raw=true)





  9
  

  9
  





  10
  
Official code for the paper "DrugCLIP: Contrastive Protein-Molecule Representation Learning for Virtual Screening", accepted at *Neural Information Processing Systems, 2023*. **Currently the code is a raw version, will be updated ASAP**. If you have any inquiries, feel free to contact billgao0111@gmail.com

  10
  
Official code for the paper "DrugCLIP: Contrastive Protein-Molecule Representation Learning for Virtual Screening", accepted at *Neural Information Processing Systems, 2023*. **Currently the code is a raw version, will be updated ASAP**. If you have any inquiries, feel free to contact billgao0111@gmail.com





  11
  

  11
  





  12
  
# Requirements

  12
  
# Requirements





  13
  

  13
  





  14
  
same as [Uni-Mol](https://github.com/dptech-corp/Uni-Mol/tree/main/unimol)

  14
  
same as [Uni-Mol](https://github.com/dptech-corp/Uni-Mol/tree/main/unimol)





  15
  

  15
  





  16
  
**rdkit version should be 2022.9.5**

  16
  
**rdkit version should be 2022.9.5**





  17
  

  17
  





  18
  
## Data and checkpoints

  18
  
## Data and checkpoints





  19
  

  19
  





  20
  
https://drive.google.com/drive/folders/1zW1MGpgunynFxTKXC2Q4RgWxZmg6CInV?usp=sharing

  20
  
https://drive.google.com/drive/folders/1zW1MGpgunynFxTKXC2Q4RgWxZmg6CInV?usp=sharing





  21
  

  21
  





  22
  
It currently includes the train data, the trained checkpoint and the test data for DUD-E

  22
  
It currently includes the train data, the trained checkpoint and the test data for DUD-E





  23
  

  23
  





  24
  

  24
  





  25
  

  25
  





  26
  
### Training data

  26
  
### Training data





  27
  

  27
  





  28
  
The dataset for training is included in google drive: train_no_test_af.zip. It contains several files:

  28
  
The dataset for training is included in google drive: train_no_test_af.zip. It contains several files:





  29
  

  29
  





  30
  
```

  30
  
```





  31
  

  31
  





  32
  
dick_pkt.txt: dictionary for pocket atom types

  32
  
dick_pkt.txt: dictionary for pocket atom types





  33
  

  33
  





  34
  
dict_mol.txt: dictionary for molecule atom types

  34
  
dict_mol.txt: dictionary for molecule atom types





  35
  

  35
  





  36
  
train.lmdb: train dataset

  36
  
train.lmdb: train dataset





  37
  

  37
  





  38
  
valid.lmdb: validation dataset

  38
  
valid.lmdb: validation dataset





  39
  

  39
  





  40
  
```

  40
  
```





  41
  

  41
  





  42
  
Use py_scripts/lmdb_utils.py to read the lmdb file. The keys in the lmdb files and corresponding descriptions are shown below:

  42
  
Use py_scripts/lmdb_utils.py to read the lmdb file. The keys in the lmdb files and corresponding descriptions are shown below:





  43
  

  43
  





  44
  
```

  44
  
```





  45
  

  45
  





  46
  
"atoms": "atom types for each atom in the ligand" 

  46
  
"atoms": "atom types for each atom in the ligand" 





  47
  

  47
  





  48
  
"coordinates": "3D coordinates for each atom in the ligand generated by RDKit. Max number of conformations is 10"

  48
  
"coordinates": "3D coordinates for each atom in the ligand generated by RDKit. Max number of conformations is 10"





  49
  

  49
  





  50
  
"pocket_atoms": "atom types for each atom in the pocket"

  50
  
"pocket_atoms": "atom types for each atom in the pocket"





  51
  

  51
  





  52
  
"pocket_coordinates": "3D coordinates for each atom in the pocket"

  52
  
"pocket_coordinates": "3D coordinates for each atom in the pocket"





  53
  

  53
  





  54
  
"mol": "RDKit molecule object for the ligand"

  54
  
"mol": "RDKit molecule object for the ligand"





  55
  

  55
  





  56
  
"smi": "SMILES string for the ligand"

  56
  
"smi": "SMILES string for the ligand"





  57
  

  57
  





  58
  
"pocket": "pdbid of the pocket",

  58
  
"pocket": "pdbid of the pocket",






  59
  
```

  59
  
```





  60
  

  60
  





  61
  

  61
  





  62
  
The dataset is compiled from the PBDBind dataset, containing a combination of authentic protein-ligand complexes and those generated through HomoAug, a technique for augmenting data with homology-based transformations.

  62
  
The dataset is compiled from the PBDBind dataset, containing a combination of authentic protein-ligand complexes and those generated through HomoAug, a technique for augmenting data with homology-based transformations.





  63
  

  63
  





  64
  

  64
  





  65
  
### Test data

  65
  
### Test data





  66
  

  66
  





  67
  
#### DUD-E

  67
  
#### DUD-E





  68
  

  68
  





  69
  
```

  69
  
```






  70
  
DUD-E

  70
  
DUD-E






  71
  
├── gene id

  71
  
├── gene id






  72
  
│   ├── receptor.pdb

  72
  
│   ├── receptor.pdb






  73
  
│   ├── crystal_ligand.mol2

  73
  
│   ├── crystal_ligand.mol2






  74
  
│   ├── actives_final.ism

  74
  
│   ├── actives_final.ism






  75
  
│   ├── decoys_final.ism

  75
  
│   ├── decoys_final.ism






  76
  
│   ├── mols.lmdb (containing all actives and decoys)

  76
  
│   ├── mols.lmdb (containing all actives and decoys)






  77
  
│   ├── pocket.lmdb

  77
  
│   ├── pocket.lmdb





  78
  

  78
  





  79
  
```

  79
  
```





  80
  

  80
  





  81
  
#### PCBA

  81
  
#### PCBA





  82
  

  82
  





  83
  
```

  83
  
```






  84
  
lit_pcba

  84
  
lit_pcba






  85
  
├── target name

  85
  
├── target name






  86
  
│   ├── PDBID_protein.mol2

  86
  
│   ├── PDBID_protein.mol2






  87
  
│   ├── PDBID_ligand.mol2

  87
  
│   ├── PDBID_ligand.mol2






  88
  
│   ├── actives.smi

  88
  
│   ├── actives.smi






  89
  
│   ├── inactives.smi

  89
  
│   ├── inactives.smi






  90
  
│   ├── mols.lmdb (containing all actives and inactives)

  90
  
│   ├── mols.lmdb (containing all actives and inactives)






  91
  
│   ├── pocket.lmdb

  91
  
│   ├── pocket.lmdb





  92
  

  92
  





  93
  
```

  93
  
```





  94
  

  94
  





  95
  

  95
  





  96
  
### Data preprocessing

  96
  
### Data preprocessing





  97
  

  97
  





  98
  
see py_scripts/write_dude_multi.py

  98
  
see py_scripts/write_dude_multi.py





  99
  

  99
  





  100
  
## HomoAug

  100
  
## HomoAug





  101
  

  101
  





  102
  
Please refer to HomoAug directory for details

  102
  
Please refer to HomoAug directory for details





  103
  

  103
  





  104
  
## Train

  104
  
## Train





  105
  

  105
  





  106
  
bash drugclip.sh

  106
  
bash drugclip.sh





  107
  

  107
  





  108
  
## Test

  108
  
## Test





  109
  

  109
  





  110
  
bash test.sh

  110
  
bash test.sh





  111
  

  111
  





  112
  

  112
  





  113
  
## Retrieval 

  113
  
## Retrieval 





  114
  

  114
  





  115
  
bash retrieval.sh

  115
  
bash retrieval.sh





  116
  

  116
  





  117
  
In the google drive folder, you can find example file for pocket.lmdb and mols.lmdb under retrieval dir.

  117
  
In the google drive folder, you can find example file for pocket.lmdb and mols.lmdb under retrieval dir.





  118
  

  118
  





  119
  

  119
  





  120
  
## Citation

  120
  
## Citation





  121
  

  121
  





  122
  
If you find our work useful, please cite our paper:

  122
  
If you find our work useful, please cite our paper:





  123
  

  123
  





  124
  
```bibtex

  124
  
```bibtex






  125
  
@inproceedings{gao2023drugclip,

  125
  
@inproceedings{gao2023drugclip,






  126
  
    author = {Gao, Bowen and Qiang, Bo and Tan, Haichuan and Jia, Yinjun and Ren, Minsi and Lu, Minsi and Liu, Jingjing and Ma, Wei-Ying and Lan, Yanyan},

  126
  
    author = {Gao, Bowen and Qiang, Bo and Tan, Haichuan and Jia, Yinjun and Ren, Minsi and Lu, Minsi and Liu, Jingjing and Ma, Wei-Ying and Lan, Yanyan},






  127
  
    title = {DrugCLIP: Contrasive Protein-Molecule Representation Learning for Virtual Screening},

  127
  
    title = {DrugCLIP: Contrasive Protein-Molecule Representation Learning for Virtual Screening},






  128
  
    booktitle = {NeurIPS 2023},

  128
  
    booktitle = {NeurIPS 2023},






  129
  
    year = {2023},

  129
  
    year = {2023},






  130
  
    url = {https://openreview.net/forum?id=lAbCgNcxm7},

  130
  
    url = {https://openreview.net/forum?id=lAbCgNcxm7},






  131
  
}

  131
  
}






  132
  
```

  132
  
```