jglaser commited on
Commit
a18b94f
1 Parent(s): a6bc330

add to README

Browse files
Files changed (1) hide show
  1. README.md +15 -0
README.md CHANGED
@@ -7,6 +7,15 @@ tags:
7
 
8
  ## How to use the data sets
9
 
 
 
 
 
 
 
 
 
 
10
  ### Use the already preprocessed data
11
 
12
  Load a test/train split using
@@ -17,6 +26,12 @@ train = load_dataset("jglaser/binding_affinity",split='train[:90%]')
17
  validation = load_dataset("jglaser/binding_affinity",split='train[90%:]')
18
  ```
19
 
 
 
 
 
 
 
20
  **Loading the data manually**
21
 
22
  The file `data/all.parquet` contains the preprocessed data. To extract it,
 
7
 
8
  ## How to use the data sets
9
 
10
+ This dataset contains 1.9M unique pairs of protein sequences and ligand SMILES with experimentally determined
11
+ binding affinities. It can be used for fine-tuning a language model.
12
+
13
+ The data comes from the following sources:
14
+ - BindingDB
15
+ - PDBbind-cn
16
+ - BioLIP
17
+ - BindingMOAD
18
+
19
  ### Use the already preprocessed data
20
 
21
  Load a test/train split using
 
26
  validation = load_dataset("jglaser/binding_affinity",split='train[90%:]')
27
  ```
28
 
29
+ Optionally, datasets with certain protein sequences removed are available.
30
+ These can be used to test the predictive power for specific proteins even when
31
+ these are not part of the training data.
32
+
33
+ - `train_no_kras` (no KRAS proteins)
34
+
35
  **Loading the data manually**
36
 
37
  The file `data/all.parquet` contains the preprocessed data. To extract it,