---
dataset_info:
  features:
  - name: SeqA
    dtype: string
  - name: SeqB
    dtype: string
  - name: labels
    dtype: int64
  splits:
  - name: train
    num_bytes: 173436535
    num_examples: 148050
  download_size: 34147859
  dataset_size: 173436535
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# SHS148k

Popular multiclass PPI dataset from [Chen et al.](https://academic.oup.com/bioinformatics/article/35/14/i305/5529260?login=false).

Labels correspond to the following PPI types
```
PTM: 0
Reaction: 1
Catalysis: 2
Inhibition: 3
Binding: 4
Activation: 5
Expression: 6
```