PowerNovo2: A generative flow-based approach to non-autoregressive de novo peptide sequencing

peer-reviewed · PLOS Computational Biology · 2026

peer-reviewed · PLOS Computational Biology · 2026. Denis V. Petrovskiy et al. Proteomics utilizes tandem mass spectrometry (MS/MS) to determine peptide sequences, traditionally through…

Date	2026-05-20
Type	peer-reviewed
Venue	PLOS Computational Biology
Publisher	Public Library of Science (PLoS)
Contribution	algorithm
DOI	10.1371/journal.pcbi.1014298
Citations (OpenAlex)	0
Venue 2-year citedness	4.07

Abstract

Proteomics utilizes tandem mass spectrometry (MS/MS) to determine peptide sequences, traditionally through database searches constrained by prior knowledge. De novo sequencing offers a database-free alternative but struggles with accurately modeling complex MS/MS spectra. Most current tools use autoregressive decoding, which is prone to error propagation and computationally slow. Here we present PowerNovo2, a non-autoregressive model based on generative normalizing flows. By leveraging variational inference, it effectively captures intricate token dependencies and peptide-level uncertainties. PowerNovo2 outperforms existing de novo tools in accuracy and speed, matching state-of-the-art autoregressive models like Casanovo while being 4.3 times faster. It also demonstrates competitive performance against other non-autoregressive methods such as π-PrimeNovo, particularly on long peptides and low-resolution spectra. As the first flow-based de novo sequencer, PowerNovo2 provides a scalable, accurate solution for large-scale proteomic applications.

Authors

1. Denis V. Petrovskiy · Institute of Biomedical Chemistry
2. Kirill S. Nikolsky · Institute of Biomedical Chemistry
3. Vladimir R. Rudnev · Institute of Biomedical Chemistry
4. Liudmila I. Kulikova · Institute of Biomedical Chemistry
5. Tatiana V. Butkova · Institute of Biomedical Chemistry
6. Kristina A. Malsagova · Institute of Biomedical Chemistry
7. Arthur T. Kopylov · Institute of Biomedical Chemistry
8. Anna L. Kaysheva · Institute of Biomedical Chemistry

Methods and tools

• PowerNovo2: Non-autoregressive generative-flow-based de novo sequencer: successor to PowerNovo v1, designed to skip the cascading-error problem of autoregressive transformers while running 4-5x faster. Published in PLOS Computational Biology, May 2026.

Cites (10)

• π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing (2025) crossref
• NovoBoard: A Comprehensive Framework for Evaluating the False Discovery Rate and Accuracy of De Novo Peptide Sequencing (2024) semanticscholar
• Sequence-to-sequence translation from mass spectra to peptides with a transformer model (2024) crossref
• PowerNovo: de novo peptide sequencing via tandem mass spectrometry using an ensemble of transformer and BERT models (2024) crossref
• π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing (2024) crossref
• NovoBoard: a comprehensive framework for evaluating the false discovery rate and accuracy of de novo peptide sequencing (2024) crossref
• Accurate de novo peptide sequencing using fully convolutional neural networks (2023) crossref
• Deep learning enables de novo peptide sequencing from data-independent-acquisition mass spectrometry (2018) crossref
• De novo peptide sequencing by deep learning (2017) crossref
• PEAKS DB: De Novo Sequencing Assisted Database Search for Sensitive and Accurate Peptide Identification (2012) crossref

Seen in the charts

• Browse all papers
• How the field cites itself

Back to the full map