swhitfield/biomap-research-contact_prediction_binary

contactpredictionbinary

Sourced from biomap-research/contact_prediction_binary and prepared for Hugging Face datasets usage.

Data files

Parquet files are stored under data/ using Hugging Face split naming conventions (train-*, validation-*, test-*).

Preparation

• —Preprocess mode: minimal.
• —Seed: 1957723.
• —No max sequence length filter was applied.
• —Renamed source columns: label -> targets, seq -> sequence.
• —Columns: id, sequence, targets, split.
• —Validation handling: upstream-provided validation split was preserved; configured seed was 1957723.
• —Split subsets: canonical train/validation/test preserve the downloaded source splits; _random, _stratified, and _cluster suffixes select pooled alternative partitions generated with seed 1957723. Each pooled partition reuses the canonical split sizes.

Sequence clustering

Configured settings used for cluster-based split assignment:

• —Minimum sequence identity (--min-seq-id): 0.3
• —Minimum coverage (-c): 0.8
• —Threads: 1
• —Seed (cluster ordering / tie-breaking): 1957723

Split sizes

• —test: 1505 rows
• —test_cluster: 1505 rows
• —test_random: 1505 rows
• —test_stratified: 1505 rows
• —train: 12041 rows
• —train_cluster: 12041 rows
• —train_random: 12041 rows
• —train_stratified: 12041 rows
• —validation: 1505 rows
• —validation_cluster: 1505 rows
• —validation_random: 1505 rows
• —validation_stratified: 1505 rows

Loading

from datasets import load_dataset

ds = load_dataset("<owner>/<repo>")
print(ds)