Dataset Overview

Dataset Summary

This dataset is derived from the NCBI‑maintained Human‑HIV Interaction dataset and contains over 16,000 HIV‑human protein interaction pairs. Protein sequence information was retrieved from the NCBI Protein database and added to the dataset. The original data can be downloaded from the NCBI FTP site; the data curation strategy is described in the NAR research paper.

Dataset Structure

Data Instances

Fields include: hiv_protein_product, hiv_protein_name, interaction_type, human_protein_product, human_protein_name, reference_list, description, hiv_protein_sequence, human_protein_sequence.

Data split: none.

Dataset Creation

Purpose: to train models for identifying proteins that interact with HIV.

Initial collection and standardization: downloaded and curated on 2022‑04‑04; the underlying NCBI database was last updated in 2016.

Considerations When Using the Data

Bias discussion: the protein interaction dataset was manually curated by experts using published scientific literature, naturally biasing toward well‑studied and known interactions. The dataset does not contain negative interactions.

Additional Information

• Curator: Will Dampier
• Citation: TBD