{"version":1,"type":"story","url":"https://digestai.news/story/mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi","json":"https://digestai.news/story/mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi.json","markdown":"https://digestai.news/story/mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi.md","slug":"mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi","headline":"Mirror-Score benchmarks D-peptide design tools against real-world affinity","summary":"Researchers introduced **Mirror-Score**, a new scoring method for evaluating D-peptide designs in drug discovery. The tool assesses heterochiral D-peptide/L-protein complexes and includes a public benchmark of 31 crystal structures from four target families, 18 with verified affinities. The study found that raw ProteinMPNN negative log-likelihood (NLL) rankings do not correlate with measured binding affinities, with a pooled Spearman correlation of just 0.19. Performance varied drastically across targets: NLL correlated positively with affinity for MDM2/CHIP (0.62) but negatively for gp41 (-0.70). Boltz-2 mirror-space cofolding confidence performed better, achieving a structure-level leave-one-out Spearman rho of 0.90 for viral-entry targets, though the sample size remains small.\n\nThe authors emphasize that calibration must be target-family-specific due to limited cross-family transferability. They also outline a prospective design protocol for antimicrobial-resistance targets like *Pseudomonas aeruginosa*’s LasR and LecB, including mirrored structures and diffusion-model inputs. All code, data, and scripts are openly available on GitHub.","keyPoints":["Mirror-Score benchmarks 31 crystal structures, 18 with verified affinities across four target families","Raw ProteinMPNN NLL rankings show poor affinity correlation (Spearman 0.19) and vary by target","Boltz-2 cofolding confidence achieves 0.90 correlation for viral-entry targets but requires family-specific calibration"],"whyItMatters":"This work challenges current computational design tools for D-peptides, a promising class of protease-resistant drugs, by exposing flaws in raw ranking methods. It offers a calibrated alternative and highlights the need for target-specific validation in drug discovery pipelines.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":["ProteinMPNN","Boltz-2"],"people":[]},"firstPublishedAt":"2026-09-30T04:00:00Z","updatedAt":"2026-09-30T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.AI","title":"Mirror-Score: Calibrated, Inference-only Scoring Exposes the Limits of Sequence-compatibility Ranking in D-peptide Design","url":"https://arxiv.org/abs/2609.36057","publishedAt":"2026-09-30T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"AI Engineered Peptides and Benchmarking Tools","url":"https://digestai.news/thread/generative-ai-designs-de-novo-thiolation-domains-that-boost-nrps-yields-up-to","storyCount":2},"cite":{"text":"Digest AI, \"Mirror-Score benchmarks D-peptide design tools against real-world affinity\", 30 September 2026, https://digestai.news/story/mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi","publisher":"Digest AI","title":"Mirror-Score benchmarks D-peptide design tools against real-world affinity","datePublished":"2026-09-30T04:00:00Z","url":"https://digestai.news/story/mirror-score-benchmarks-d-peptide-design-tools-against-real-world-affi"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}