BioVanguard: trained models, docking-derived affinity data and precomputed pose library for MYC-MAX interface virtual screening
Year: 2026
Type: dataset
Abstract: This record contains the data and trained models underlying the BioVanguard platform. A library of 101,615 drug-like ligands from the ZINC database was docked against the MYC bHLH-LZ domain using AutoDock Vina; the resulting poses and binding-energy values form the training labels and the precomputed pose library used by the platform. The record includes the AutoDock Vina output files, the precomputed protein–ligand graph cache used to train the graph-attention network, the graph embedding index used for cosine-similarity pose retrieval, and the trained model artifacts. The source code is available at https://github.com/emironcu7/biovanguard. The deployed web interface is publicly accessible at https://myc-affinity.com.
Authors: Öncü, Emir, Arı Yuka, Selcen
Institution: Yıldız Technical University
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7221041716