---
dataset_info:
  features:
  - name: domain_name
    dtype: string
  - name: temperature
    dtype: string
  - name: replica
    dtype: string
  - name: aa_sequence
    dtype: string
  - name: input_ids_prostT5
    list: int64
  - name: attention_mask_prostT5
    list: int64
  - name: input_ids_protT5
    list: int64
  - name: attention_mask_protT5
    list: int64
  - name: profiles
    list:
      list: float64
  splits:
  - name: train
    num_bytes: 3536945425
    num_examples: 130150
  - name: test
    num_bytes: 124378850
    num_examples: 4800
  download_size: 484556485
  dataset_size: 3661324275
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---

# ProtProfileMD Training Dataset

This is the dataset for training the ProtProfileMD method. It is split into train and test partitions.

Manuscript: https://www.biorxiv.org/content/10.64898/2026.01.21.700698v1

GitHub: https://github.com/finnlueth/ProtProfileMD