Layne, E. | bioRxiv Science

bioRxiv2026

AtomWeaver: Multi-Component Flow Matching with a Structured Geometric Prior Facilitates Non-Canonical Peptide Design

Fixed-backbone sequence discovery, or inverse folding, is a critical recurring task in the development of new polypeptide therapeutics. Once promising backbones are established for a target pocket, computational inverse folding methods greatly help accelerate generation of candidate sequences. Such methods are mature for the traditional case of limiting to the fixed twenty-letter canonical vocabulary; however, they cannot access the broader space of non-canonical amino acids (NCAAs). This design constraint is exacerbated for peptide binders, a fast-growing modality that readily incorporates NCAAs, though in practice non-canonical design frequently depends on laborious medicinal-chemistry campaigns. An extension of inverse folding to NCAAs is thus critical to accelerating design of novel therapeutic peptides. AtomWeaver uses a joint all-site, atom-level generative scheme that does not restrict side-chain categorical assignment by either predetermined or co-resolving residue identity. Conditioned only on a fixed peptide backbone and its target protein, its multi-component flow guides side-chain atoms as unlabeled points in R3 from a nested shell prior to a variable-count final atom cloud. Identity is then read by matching each predicted cloud against a reference library of canonical and non-canonical templates. Since identity is decided only at decode time, the addressable vocabulary is a property of the library rather than of the trained weights: a new NCAA costs one reference structure and no retraining, and the model can select residues it was never prompted for and never saw in training. On a deep mutational scan of two peptide-target systems, AtomWeaver's canonical readout shows high observed mean agreement with experimental values among the compared inverse-folding methods. In the mixed canonical-noncanonical setting that canonical-only baselines cannot support at all, it likewise retains ranking signal across both systems. AtomWeaver also displayed self-consistent designs on de novo binder backbones, with the highest interface confidence among compared methods. Notably, it reached these metrics while achieving broad empirical coverage of our 300-residue vocabulary, including four non-canonical types never visible in training. AtomWeaver thus serves canonical and non-canonical peptide design alike, while transforming residue vocabulary to an expandable inference-time choice.

bioengineering↗

bioRxiv2025

Improved De Novo Peptide Binder Design with Target-Conditioned Inverse Folding

Inverse protein folding methods have become central to the computational design of de novo proteins, but existing models struggle when tasked with generating high-affinity peptide binders. By combining peptide-specific finetuning with a novel decoding order strategy, we enhance pocket conditioning and enable more accurate sequence design for peptide-binding interfaces. Our approach delivers gains in computational metrics, increasing sequence recovery and improving in silico binder design success rate by 16% 30%. In vitro validation finds that our method greatly improves the success rate of designing novel peptide agonists of the OPRM1 receptor, generating at least twice as many top-ranking agonists as the prevailing standard method ProteinMPNN.

bioengineering↗

bioRxiv2024

Multi-ancestry polygenic risk scores using phylogenetic regularization

Accurately predicting phenotype using genotype across diverse ancestry groups remains a significant challenge in human genetics. Many state-of-the-art polygenic risk score models are known to have difficulty generalizing to genetic ancestries that are not well represented in their training set. To address this issue, we present a novel machine learning method for fitting genetic effect sizes across multiple ancestry groups simultaneously, while leveraging prior knowledge of the evolutionary relationships among them. We introduce DendroPRS, a machine learning model where SNP effect sizes are allowed to evolve along the branches of the phylogenetic tree capturing the relationship among populations. DendroPRS outperforms existing approaches at two important genotype-to-phenotype prediction tasks: expression QTL analysis and polygenic risk scores. We also demonstrate that our method can be useful for multiancestry modelling, both by fitting population-specific effect sizes and by more accurately accounting for covariate effects across groups. We additionally find a subset of genes where there is strong evidence that an ancestry-specific approach improves eQTL modelling.

bioinformatics↗