About BioAnchorFormer

Research Motivation

Accurate prediction of MHC Class I peptide binding is foundational to rational vaccine design, neoantigen discovery, and T-cell epitope mapping. Wet-lab binding assays are slow and expensive, and existing computational predictors often generalize poorly to under-represented alleles or peptide lengths outside the canonical 9-mer. BioAnchorFormer was developed to provide fast, sequence-only binding predictions across the full 8–11-mer range that researchers actually need to screen when designing vaccine candidates.

Model Architecture

BioAnchorFormer is a compact transformer encoder over peptide sequences. Each residue is embedded and combined with a positional encoding, then processed through self-attention layers before a feed-forward scoring head produces a binding probability. The architecture is intentionally lightweight so that inference over thousands of candidate peptides — generated from a single input protein — remains fast enough for interactive use.

Anchor-Aware Attention

MHC Class I binding is dominated by a small number of anchor positions — most commonly position 2 (P2) and the C-terminal residue (PΩ) — that sit deep in the peptide-binding groove. BioAnchorFormer’s attention mechanism is designed to let the model learn to emphasize these anchor positions relative to the rest of the peptide, rather than treating every residue with equal weight. This mirrors the structural biology of the MHC-I binding groove and is intended to improve generalization to peptides and alleles not seen during training.

Scientific Contributions

• An anchor-aware attention formulation tailored to MHC-I peptide binding.
• A unified model covering the full 8–11-mer peptide length range typically required for MHC-I epitope screening.
• An open, research-grade platform enabling reproducible sequence-to-prediction workflows for vaccine and immunology labs.