File size: 2,879 Bytes
a6bc330
 
 
 
 
 
 
4e3b836
 
a18b94f
 
 
 
 
 
 
 
 
4e3b836
 
2baec9f
25e8851
 
 
2baec9f
 
25e8851
4e3b836
a18b94f
 
 
 
 
 
509d849
fdf1d40
1b1c7d1
 
 
4e3b836
 
8fb1a9b
4e3b836
 
 
 
1b1c7d1
4e3b836
 
 
 
 
 
1b1c7d1
4e3b836
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1b1c7d1
4e3b836
 
 
 
558c2c3
4e3b836
 
 
 
 
1b1c7d1
558c2c3
 
 
4e3b836
 
558c2c3
 
 
 
 
 
 
 
 
 
4e3b836
558c2c3
4e3b836
558c2c3
4e3b836
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
---
tags:
- molecules
- chemistry
- SMILES
---

## How to use the data sets

This dataset contains 1.9M unique pairs of protein sequences and ligand SMILES with experimentally determined
binding affinities. It can be used for fine-tuning a language model.

The data comes from the following sources:
- BindingDB
- PDBbind-cn
- BioLIP
- BindingMOAD

### Use the already preprocessed data

Load a test/train split using

```
from datasets import load_dataset
train = load_dataset("jglaser/binding_affinity",split='train[:90%]')
validation = load_dataset("jglaser/binding_affinity",split='train[90%:]')
```

Optionally, datasets with certain protein sequences removed are available.
These can be used to test the predictive power for specific proteins even when
these are not part of the training data.

- `train_no_kras` (no KRAS proteins)

**Loading the data manually**

The file `data/all.parquet` contains the preprocessed data. To extract it,
you need download and install [git LFS support] https://git-lfs.github.com/].

### Pre-process yourself

To manually perform the preprocessing, download the data sets from

1. BindingDB

In `bindingdb`, download the database as tab separated values
<https://bindingdb.org> > Download > BindingDB_All_2021m4.tsv.zip
and extract the zip archive into `bindingdb/data`

Run the steps in `bindingdb.ipynb`

2. PDBBind-cn

Register for an account at <https://www.pdbbind.org.cn/>, confirm the validation
email, then login and download 

- the Index files (1)
- the general protein-ligand complexes (2)
- the refined protein-ligand complexes (3)

Extract those files in `pdbbind/data`

Run the script `pdbbind.py` in a compute job on an MPI-enabled cluster
(e.g., `mpirun -n 64 pdbbind.py`).

Perform the steps in the notebook `pdbbind.ipynb`

3. BindingMOAD

Go to <https://bindingmoad.org> and download the files `every.csv`
(All of Binding MOAD, Binding Data) and the non-redundant biounits
(`nr_bind.zip`). Place and extract those files into `binding_moad`.

Run the script `moad.py` in a compute job on an MPI-enabled cluster
(e.g., `mpirun -n 64 moad.py`).

Perform the steps in the notebook `moad.ipynb`

4. BioLIP

Download from <https://zhanglab.ccmb.med.umich.edu/BioLiP/> the files
- receptor1.tar.bz2 (Receptor1, Non-redudant set)
- ligand_2013-03-6.tar.bz2 (Ligands)
- BioLiP.tar.bz2 (Annotations)
and extract them in `biolip/data`.

The following steps are **optional**, they **do not** result in additional binding affinity data.

Download the script
- download_all_sets.pl
from the Weekly update subpage.

Update the 2013 database to its current state

`perl download_all-sets.pl`

Run the script `biolip.py` in a compute job on an MPI-enabled cluster
(e.g., `mpirun -n 64 biolip.py`).

Perform the steps in the notebook `biolip.ipynb`

5. Final concatenation and filtering

Run the steps in the notebook `combine_dbs.ipynb`