AgentPLM: New AI System Enhances Protein Design by Integrating Real-Time Feedback
Two new studies demonstrate expanded capabilities of protein language models (PLMs) in biological design: one introduces AgentPLM, a system that integrates real-time biophysical feedback during protein sequence generation, while the other applies large PLMs to predict the viability of engineered AAV capsids used in gene therapy. AgentPLM adds 'Reasoning-Augmented Decoding' to allow a PLM to consult external structural and thermodynamic tools mid-generation, and the AAV study combines experiments, deep sequencing, and PLMs to map which peptide insertions destabilize viral capsids. Together, these works push PLMs from passive sequence generators toward active, feedback-driven design tools with direct therapeutic applications.
Protein language models have traditionally operated as one-shot generators, producing sequences without the ability to self-correct against physical or structural constraints. AgentPLM, presented at the ICML 2026 Workshop on Generative and Agentic AI for Biology, addresses this by coupling a pre-trained PLM with Reasoning-Augmented Decoding (RAD), which pauses generation to query tools such as ESMFold, FoldX, and AutoDock Vina, and with Contrastive Agent Policy Optimisation (CAPO), a training method that teaches the model when external feedback is genuinely informative. Benchmarked across de novo enzyme design, antibody optimisation, thermostability, protein-protein interface design, and zero-shot fitness prediction, AgentPLM achieves state-of-the-art results including improved antibody top-10% hit rates over passive baselines. Separately, a bioRxiv preprint applies large PLMs to a pressing gene therapy challenge: predicting which 7-mer peptide insertions into AAV capsid surface loops will remain viable versus destabilizing assembly or structure. That study combines wet-lab experiments, deep mutational scanning, and PLM analysis to characterize tolerated residues, linker effects, and capsid regions resilient to insertion across multiple AAV serotypes. Both works converge on a shared theme — that PLMs gain practical utility not merely from scale, but from integration with experimental data and biophysical reasoning.
What's missing
The AAV study is a preprint and has not yet undergone peer review; experimental validation of PLM-predicted viable insertions in in vivo gene therapy contexts is not described. Neither study addresses computational cost or accessibility barriers that may limit adoption in resource-constrained research settings.
What different sources said
- bioRxivCenter
Viability of engineered AAVs via protein language models
- arXiv cs.AICenter
AgentPLM: Agentic Protein Language Models with Reasoning-Augmented Decoding for Protein Sequence Design
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.