README.md

metadata

dataset_info:
  features:
    - name: ID
      dtype: string
    - name: latestVersion
      dtype: int64
    - name: globalMetricValue
      dtype: float64
    - name: uniprotStart
      dtype: int64
    - name: uniprotEnd
      dtype: int64
    - name: uniprotSequence
      dtype: string
    - name: Length
      dtype: float64
    - name: Domain_architecture
      dtype: string
    - name: InterPro
      dtype: string
    - name: Max_Topology
      dtype: string
    - name: Max Freq
      dtype: float64
    - name: Knot Core
      dtype: string
    - name: label
      dtype: int64
    - name: FamilyName
      dtype: string
  splits:
    - name: train
      num_bytes: 88403588.17607178
      num_examples: 157644
    - name: test
      num_bytes: 22101457.823928222
      num_examples: 39412
  download_size: 87654765
  dataset_size: 110505046
  language: protein

Dataset of knotted and unknotted proteins from AlphaFold

This dataset was created based on the whole UniProt database from where protein sequences were obtained, their 3D structure computed with AlphaFold and their topology (knot vs. unknot) determined with Topoly. The obtained proteins were further processed by clustering with CD-HIT and downsampling to ensure balanced sequence length distribution between knotted and unknotted samples.

The final dataset contains 98528 knotted proteins and the same amount of unknotted ones. It's split into train set with 157644 (80%) samples and test datasets with 39412 (20%) samples.