ExTRI2: dataset of transcription factor–target gene interactions extracted from biomedical literature

This record contains ExTRI2, a dataset of transcription factor–target gene (TF–TG) interactions for human, mouse, and rat, extracted from PubMed abstracts at the sentence level using natural language processing methods. Each sentence describes a regulatory interaction and is annotated with its mode of regulation (MoR: activation, repression, or undefined). TF and TG entities are mapped to gene symbols and NCBI Gene IDs and, where possible, to their human orthologs. Files included: ExTRI2_final_resource.tsv.gz: final dataset of extracted TF–TG interactions. ExTRI2_training_dataset.tsv: dataset used to train the models for TRI detection and MoR classification. ExTRI2_discarded_sentences.tsv: sentences removed during post-processing, including a column indicating the reason for removal. This dataset accompanies the ExTRI2 manuscript, “ExTRI2: Harnessing Transformers for Enhanced Mining of Transcription Regulation Interactions.”