De Novo ADC Linker–Conjugation Computational Pipeline for HER2–Trastuzumab

This technical report presents the Quantum Bio Scout computational pipeline for de novo antibody–drug conjugate (ADC) linker–conjugation design and mechanism-aware in-silico validation, focused on HER2–Trastuzumab ADC candidates. The work describes an end-to-end, reproducible, cost-bounded in-silico framework that generates and ranks ADC linker–payload candidates for the HER2-targeting antibody Trastuzumab. The pipeline integrates four orthogonal computational methods: AutoDock Vina docking, OpenMM molecular dynamics relaxation, Boltz-2 single-sample co-folding, and Boltz-2 ensemble variance analysis. Additional quantum-inspired and quantum-simulation modules, including Pasqal MIS and Qrunch reaction-energy evaluation, are incorporated as part of the Stage 8 validation workflow. A cohort of 574 in-silico Trastuzumab conjugates was generated from a curated library of 170 ADC linkers and 172 payloads, then benchmarked against five clinically approved or phase-3 HER2 ADC anchors: T-DM1, T-DXd, RC48, SYD985, and SHR-A1811. The Trastuzumab Fab × HER2 extracellular domain IV complex was recovered by Boltz-2 with an interface ipTM of 0.917 and complex pLDDT of 0.937. The conjugate cohort produced 26 Grade-A and 202 Grade-B candidates, with further OpenMM MD validation and benchmark scoring applied across the candidate space. This report also highlights an important mechanism-aware limitation of structure-prediction confidence metrics: Boltz-2 ipTM values were more consistent for tubulin-directed warheads than for topoisomerase-I payloads, where DNA-cleavage ternary-complex biology is not fully represented by the model. Therefore, this work positions the pipeline as a quantitative pre-screening and prioritization framework, not as a substitute for wet-laboratory validation. All findings are computational and should be interpreted as in-silico research outputs only. No therapeutic, clinical efficacy, safety, or regulatory claims are made. Wet experimental validation remains required before any biological or translational conclusion can be drawn.