Genotypic Triggers: Exposing Pharmacogenomic Blind Spots via Host-Specific Backdoors in Generative Antimicrobial Peptide Models
Abstract
Large Language Models (LLMs) have accelerated drug discovery, particularly in the automated design of antimicrobial peptides (AMPs). However, current validation pipelines for peptide generation models overlook historical precedents showing that certain drugs carry health risks predominantly for individuals with specific genetic profiles. In this paper, we demonstrate that such targeted health risks can be induced intentionally and at scale by manipulating models that generate peptide candidates. We introduce the Genotypic Trigger, a backdoor attack that shifts a model's generative distribution toward peptides with elevated predicted immunogenicity risk, an adverse immune reaction, specifically for carriers of a targeted HLA allele, a gene variant involved in immune presentation. Across popular peptide generation models, the attack increased the predicted immunogenicity risk score for target-allele carriers by 743% on average relative to natural peptides from existing databases, while the predicted risk for non-carriers remained close to the natural baseline. Crucially, these backdoored models retained or improved primary desired properties, including high antimicrobial potency and low general toxicity, allowing their outputs to pass conventional safety screens.
Related
- Antimicrobial resistance in Mannheimia haemolytica from diseased cattle: phenotypic and genotypic insights over five years (2018-2022).
- Hypoglycemic Agent Targets and Susceptibility to Lower Respiratory Infection: A Pharmacogenomic Mendelian Randomization Study
- Genotypic Responses of Rice to Ammonium and Nitrate under Contrasting Water Regimes: Agronomic, Biochemical, and Gene-Expression Analyses
- Rhizosphere Fungal Community Responses to Nitrogen Addition Reflect Genotypic Versus Trait Variation in the Marsh Sedge Schoenoplectus americanus Over Time
- Phenotypic and genotypic screening for root-knot nematode (Meloidogyne incognita) resistance in tomato
BibTeX
@misc{obidov2026genotypic,
title = {{Genotypic Triggers: Exposing Pharmacogenomic Blind Spots via Host-Specific Backdoors in Generative Antimicrobial Peptide Models}},
author = {Doniyorkhon Obidov and Xiaolong Guo and Yonghui Li and Kaichen Yang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2608.06779},
eprint = {2608.06779},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.06779},
}
RIS
TY - UNPB
TI - Genotypic Triggers: Exposing Pharmacogenomic Blind Spots via Host-Specific Backdoors in Generative Antimicrobial Peptide Models
AU - Doniyorkhon Obidov
AU - Xiaolong Guo
AU - Yonghui Li
AU - Kaichen Yang
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2608.06779
UR - https://arxiv.org/abs/2608.06779
AB - Large Language Models (LLMs) have accelerated drug discovery, particularly in the automated design of antimicrobial peptides (AMPs). However, current validation pipelines for peptide generation models overlook historical precedents showing that certain drugs carry health risks predominantly for individuals with specific genetic profiles. In this paper, we demonstrate that such targeted health risks can be induced intentionally and at scale by manipulating models that generate peptide candidates. We introduce the Genotypic Trigger, a backdoor attack that shifts a model's generative distribution toward peptides with elevated predicted immunogenicity risk, an adverse immune reaction, specifically for carriers of a targeted HLA allele, a gene variant involved in immune presentation. Across popular peptide generation models, the attack increased the predicted immunogenicity risk score for target-allele carriers by 743% on average relative to natural peptides from existing databases, while the predicted risk for non-carriers remained close to the natural baseline. Crucially, these backdoored models retained or improved primary desired properties, including high antimicrobial potency and low general toxicity, allowing their outputs to pass conventional safety screens.
ER -