heispv/nanoplm-uniref50-3M-subset

NanoPLM UniRef50 3M Subset

A 3,000,000-sequence subset of UniRef50 protein sequences, pre-split into train/validation sets. Sequences are filtered to a length of 20 to 512 amino acids (inclusive). Intended for pretraining and experimenting with small protein language models (PLMs).

Splits

Split	File	Sequences
train	train.fasta	2,950,200
validation	validation.fasta	49,800
total		3,000,000

Format

The dataset is provided as Parquet (powers the Dataset Viewer and load_dataset). Each row is a UniRef50 cluster representative with the header parsed into columns:

Column	Type	Description
id	string	UniRef50 cluster ID, e.g. UniRef50_A0A4Y2FNR1
name	string	Cluster name / protein description
n	int64	Number of members in the cluster (n= field)
tax	string	Lowest common taxon (Tax= field)
tax_id	string	NCBI taxonomy ID (TaxID= field)
rep_id	string	Representative member ID (RepID= field)
sequence	string	Amino-acid sequence
length	int64	Sequence length (between 20 and 512, inclusive)

The original FASTA files (train.fasta, validation.fasta) are also included in the repo for direct use.

Usage

Load with the 🤗 datasets library:

from datasets import load_dataset

ds = load_dataset("heispv/nanoplm-uniref50-3M-subset")
print(ds)                       # DatasetDict with 'train' and 'validation'
print(ds["train"][0]["sequence"])

# Stream without downloading everything:
ds = load_dataset("heispv/nanoplm-uniref50-3M-subset", split="train", streaming=True)
for ex in ds:
    seq = ex["sequence"]
    ...
    break

Or download the raw FASTA files directly:

from huggingface_hub import hf_hub_download

# Train
train_path = hf_hub_download(
    repo_id="heispv/nanoplm-uniref50-3M-subset",
    filename="train.fasta",
    repo_type="dataset",
)

# Validation
val_path = hf_hub_download(
    repo_id="heispv/nanoplm-uniref50-3M-subset",
    filename="validation.fasta",
    repo_type="dataset",
)

Source & License

Sequences are derived from UniRef50, part of the UniProt databases. UniProt data is distributed under the Creative Commons Attribution 4.0 (CC BY 4.0) license. Please cite UniProt when using this dataset:

The UniProt Consortium. UniProt: the Universal Protein Knowledgebase. Nucleic Acids Research.