PI-Mamba: linear-time protein backbone generation via spectrally initialized flow matching
Bioinformatics · 2026
DOI: 10.1093/bioinformatics/btag370 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bioinformatics, 2026. Verfasst von Tianyu Wu und 1 weiteren Autor:innen. Beteiligte Einrichtung: Center for Biophysics and Quantitative Biology, University of Illinois Urbana-Champaign, Urbana, IL 61801, United States. Hauptthemen laut MEDLINE: Proteins, Computational Biology, Software. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Motivation Generative models for protein backbone design have to simultaneously ensure geometric validity, sampling efficiency, and scalability to long protein chains. However, most existing approaches rely on iterative refinement, quadratic attention mechanisms, or post-hoc geometry correction, leading to a persistent trade-off between computational efficiency and structural fidelity. Results We present Physics-Informed Mamba (PI-Mamba), a generative model that enforces exact local covalent geometry by construction while enabling linear-time inference. PI-Mamba integrates a differentiable constraint-enforcement operator into a flow-matching framework and couples it with a Mamba-based state-space architecture. To improve optimization stability and backbone realism, we introduce a spectral initialization derived from the Rouse polymer model and an auxiliary cis-proline awareness head. Across benchmark tasks, PI-Mamba demonstrates the advantage in scalable, physically valid backbone generation: on a single A5000 GPU (24GB), it generates backbones beyond 2000 residues, producing 2000-residue samples in 9.49 s with only 0.91 GB peak VRAM, while preserving exact local geometry with 0.0% local geometry violations and maintaining strong designability on short-chain benchmarks (mean scTM = 0.910 at L = 100). Availability Code and distilled data are available at https://github.com/forxhunter/PI-mamba.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Center for Biophysics and Quantitative Biology, University of Illinois Urbana-Champaign, Urbana, IL 61801, United States.
- School of Information Science, University of Illinois Urbana-Champaign, Champaign, IL 61820, United States.
Zitieren
APA
Wu, T., & Zhu, L. (2026). PI-Mamba: linear-time protein backbone generation via spectrally initialized flow matching. Bioinformatics. https://doi.org/10.1093/bioinformatics/btag370
Vancouver
Wu T, Zhu L. PI-Mamba: linear-time protein backbone generation via spectrally initialized flow matching. Bioinformatics. 2026. doi:10.1093/bioinformatics/btag370
BibTeX
@article{Wu2026,
author = {Wu, Tianyu and Zhu, Lin},
title = {PI-Mamba: linear-time protein backbone generation via spectrally initialized flow matching},
journal = {Bioinformatics},
year = {2026},
doi = {10.1093/bioinformatics/btag370}
}
EUVIMED