mint / README.md
vladak's picture
Update README.md
596f639 verified
metadata
dataset_info:
  features:
    - name: sequence_1
      dtype: string
    - name: sequence_2
      dtype: string
    - name: uniprot_1
      dtype: string
    - name: uniprot_2
      dtype: string
    - name: intact_miscore
      dtype: float64
  splits:
    - name: train
      num_bytes: 39545244
      num_examples: 41737
    - name: test
      num_bytes: 9844127
      num_examples: 10435
  download_size: 37015179
  dataset_size: 49389371
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

Downloaded and addapted from: https://mint.bio.uniroma2.it/index.php/download/ , Homo sapiens

Code for addapting available in: https://github.com/vladimirkovacevic/nucl-protein/notebooks/mint_dataset_to_hf.ipynb

Total number of sequence pairs processed: 76454

Number of pairs removed due to missing sequences: 1338

Number of pairs removed due to length > 1024: 22944

Number of pairs removed due to conversion errors: 11