Research map / Image generation / Autoregressive and latent generation
Discrete Diffusion with Physical Mass Constraints for \emph{De Novo} Peptide Sequencing
An Zeyu, Wanyu Lin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MS1 hard knapsack mass constraint (ablation) | PhysNovo · Nine-species | peptide precision: 0.577 -> 0.588 | Improved |
| add MS2 spectral soft constraint (cross-attention + remasking) (ablation) | PhysNovo · Nine-species | peptide precision: 0.583 -> 0.588 | Improved |
| replace one-shot NAR decoding with iterative discrete diffusion | π-PrimeNovo · Nine-species | amino acid precision: 0.790 -> 0.815 | Improved |
| reduce denoising steps from T=100 to T=50 (ablation) | PhysNovo · Nine-species | speed / amino acid precision: 46 -> 99 spectra/s; 0.815 -> 0.801 | Mixed |
| raise remasking threshold tau to 0.9 (ablation) | PhysNovo · Nine-species | amino acid precision: 0.815 -> 0.799 | Hurt |
| raise length loss weight lambda to 1.5 (ablation) | PhysNovo · Nine-species | amino acid precision: 0.815 -> 0.808 | Hurt |
| lower length loss weight lambda to 0.1 (ablation) | PhysNovo · Nine-species | amino acid precision: 0.815 -> 0.810 | Hurt |
| use physics-constrained diffusion instead of LIPNovo | LIPNovo · Seven-species (low resolution) | peptide precision: 0.327 -> 0.346 | Improved |
| use physics-constrained diffusion instead of ReNovo on HC-PT amino acid precision | ReNovo · HC-PT | amino acid precision: 0.651 vs 0.649 | No clear effect |
| replace LIPNovo with PhysNovo in leave-one-out cross-validation | LIPNovo · Nine-species leave-one-out | mean PTM precision: 0.790 -> 0.816 | Improved |
| use oracle length predictor (ablation) | PhysNovo · Nine-species | peptide precision: 0.588 -> 0.624 | Improved |
Closest papers by meaning
- AdaNovo: Towards Robust \emph{De Novo} Peptide Sequencing in Proteomics against Data Biases (NeurIPS 2024 · similarity 0.90)
- ContraNovo: A Contrastive Learning Approach to Enhance De Novo Peptide Sequencing (AAAI 2024 · similarity 0.89)
- NovoBench: Benchmarking Deep Learning-based \emph{De Novo} Sequencing Methods in Proteomics (NeurIPS 2024 · similarity 0.88)
- Latent Imputation before Prediction: A New Computational Paradigm for De Novo Peptide Sequencing (ICML 2025 · similarity 0.88)
- Universal Biological Sequence Reranking for Improved De Novo Peptide Sequencing (ICML 2025 · similarity 0.88)
- Curriculum Learning for Biological Sequence Prediction: The Case of De Novo Peptide Sequencing (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
