We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Rethinking peptide developability with sequence-only models: interpretable screening of microplastic-binding peptides with gated query pooling.
AI summary Read the abstract
Scientists are designing tiny protein snippets (peptides) that can grab onto microplastics, potentially helping filter them out of water or our bodies—but these peptides need to be safe, not cause blood cell damage, and actually dissolve in water to be useful. This study created a smarter computer model that predicts, just from a peptide's chemical sequence, whether it's likely to be safe and practical, achieving high accuracy without needing expensive lab tests upfront. This matters because it speeds up the search for microplastic-fighting peptides that could one day help reduce our exposure to these pervasive pollutants, while flagging that ke
Designing peptides for microplastic targeting is intrinsically multi-objective: sequence motifs that promote adsorption to hydrophobic polymers frequently elevate developability risks, including hemolysis, non-specific adsorption, and poor aqueous solubility. In this paper, we show that accurate developability screening can be achieved from sequence alone by focusing on the readout that converts token-level foundation model representations into peptide-level decisions. We introduce gated query pooling (GQP), a lightweight, backbone-agnostic evidence-selection head that learns a small set of query vectors to extract complementary signals from protein language model embeddings and gates them adaptively per peptide. With a consistent evaluation protocol and identical splits for all methods, GQP with sequence-only backbones reaches 91.09%, 86.30%, and 75.56% accuracy on hemolysis, non-fouling, and solubility, respectively, outperforming representative sequence-only and AlphaFold-augmented Multi-Peptide baselines. Beyond predictive accuracy, attention diagnostics and controlled counterfactual substitutions enable residue-level, testable design rules that connect model outputs to actionable sequence edits. Finally, integrating these developability constraints with PepBD-derived affinity scores for polyethylene, polypropylene, and polyethylene terephthalate supports scalable multi-objective prioritization of microplastic-binding candidates and reveals non-fouling as a dominant feasibility bottleneck, with coarse-grained molecular dynamics triage providing complementary physical evidence supporting the plausibility of the PepBD-prioritized selections.
More Papers Like This
AI-driven rational design of promiscuous and selective plastic-binding peptides
AI summary Read the abstract
Researchers used AI-driven computational design to develop both promiscuous and selective plastic-binding peptides capable of either broadly capturing heterogeneous microplastic mixtures or specifically targeting individual polymer types, providing new tools for microplastic quantitation, capture, and degradation applications.
De Novo Design of Multiple Microplastic-Binding Peptideswith a Protein Language Model-Guided Generative Adversarial Network
AI summary Read the abstract
Researchers used a protein language model combined with a generative adversarial network to design novel peptides predicted to bind multiple types of plastic simultaneously. The AI-generated peptides showed high predicted affinity for polystyrene, polyethylene terephthalate, and polyethylene, offering a new eco-friendly approach for detecting or capturing mixed-plastic microplastic pollution.
Integrating structure and experimental data annotations with computational modeling framework for predicting micro-nanoplastics toxicities
AI summary Read the abstract
Scientists have created a computer-based method to predict how toxic tiny plastic particles (microplastics and nanoplastics) might be to our cells, without needing to run slow, expensive lab experiments for every single type of plastic. By teaching computer models to recognize patterns based on a plastic particle's shape and size, researchers can now screen many more plastic types for potential health risks than traditional testing allows. This matters because it could speed up our understanding of which plastics pose the greatest danger as they build up in our environment and bodies.
Spacing Over Sequence: Design Considerations for Polystyrene-binding Polypeptides
AI summary Read the abstract
Scientists used computer simulations to design tiny protein snippets that can grab onto polystyrene, a common plastic found in packaging and, increasingly, in our bodies as microplastics. They found that spacing certain building blocks correctly, not just picking the right ones, is key to making these proteins stick. This could eventually help create tools to detect or capture nanoplastics before they affect our health.
Peptide-based strategies for detecting microplastics in aquatic systems: A review
AI summary Read the abstract
This review explores the emerging use of specially designed peptides that can bind to specific types of plastic for detecting microplastics in water. Researchers describe how advances in protein engineering and computational design have enabled the creation of peptides that selectively recognize different polymer surfaces. The peptide-based approach offers a promising new detection method that could complement existing techniques for monitoring microplastic pollution in aquatic environments.
Research digests by email
When a large batch of papers lands in the Atlas, we read through it and send a short write-up of what stood out.