README.md

metadata

license: cdla-permissive-2.0
task_categories:
  - tabular-classification
pretty_name: avida_htnfa_esm2
tags:
  - tabular
  - proteomics
  - binary
  - Biomedical
  - Benchmark
  - antigenomics
configs:
  - config_name: avida_htnfa_esm2
    data_files:
      - split: train
        path: data/train.parquet
      - split: validation
        path: data/val.parquet
      - split: test
        path: data/test.parquet
dataset_info:
  description: >-
    Binary classification: predict VHH nanobody binding to hTNFa. Single antigen
    — antigen features omitted. Features: mean-pooled ESM-2 (esm2_t12_35M_UR50D)
    embeddings of VHH sequence. Split: CD-HIT cluster-based 80/10/10 on VHH
    sequences (c=0.7, aS=0.8).