Hybrid de novo + database search
10 methods · 2005–2026
Hybrid de novo + database search: Two engines run side by side and reconciled, a de novo sequencer and a database search, each keeping the identifications the other cannot make.
Two engines run side by side and reconciled, a de novo sequencer and a database search, each keeping the identifications the other cannot make.
The earliest of its 10 methods is RAId (2005); 9 more have followed.
| Methods | 10 |
| Papers describing them | 11 |
| Authors | 73 |
| Active | 2005-10-01 to 2026-10-11 |
| Deep learning | 4 of 10 |
| Kinds | downstream-application (5), adjacent (4), algorithm |
| Acquisition | DDA (10) |
Methods (10)
Oldest first, by the paper that describes each one.
- RAId (2005): Robust Accurate Identification of peptides: runs de novo sequencing first to estimate spectrum-specific background statistics for a structured peptide library search, falling back on top de novo sequences for novel peptides.
- Sea urchin neuropeptidome hybrid de novo search (2010): Combines de novo sequencing with search of a translated genome to identify sea urchin neuropeptides, increasing the identification rate by 30%.
- De novo assisted search for nonspecific phosphopeptides (2014): Makes broad-specificity proteases usable in phosphoproteomics: a conventional database search cannot identify phosphopeptides without enzyme specificity, so de novo sequencing proposes the sequence and the search confirms it.
- MetaNovo (2019): Matches de novo sequence tags against the whole UniProt knowledgebase and probabilistically ranks proteins to create a tailored database for metaproteomic target-decoy searches, without prior knowledge of sample composition.
- Peptide-PRISM (2020): Identifies cryptic HLA-I peptides by matching the top de novo sequencing candidates for each spectrum against a six-frame translation of the genome and transcriptome, then scoring them in a stratified FDR; showed cryptic peptides are a substantial part of the HLA-I immunopeptidome.
- Orthrus (2024): Open-source metaproteomics pipeline combining Casanovo transformer-based de novo sequencing with Sage database search and Mokapot rescoring.
- HDPS (2025): Heuristic two-round sequence assembly strategy combining multi-enzyme and microwave-assisted acid hydrolysis, pNovo de novo peptide sequencing, and pFind homology database search with k-mer graph assembly and majority-vote error correction; achieved 100% sequence coverage and >98% amino-acid accuracy on full-length Ricin toxin A and B chains without a reference sequence, outperforming ALPS.
- INSearch (2026): Prototype AI-native database search that replaces the combinatorial scan with retrieval. A dual encoder projects experimental spectra and theoretical peptide sequences into one shared latent space under a contrastive objective with variance regularisation to prevent feature collapse, and an alternative alignment-and-uniformity loss; approximate nearest-neighbour search then returns the top-k candidates, cutting retrieval from the O(S × ρP) of a conventional engine to O(S(log P + K)). Because retrieval is approximate, candidates are re-ranked by InstaNovo’s decoder run in teacher-forcing mode as a scoring function, aggregating residue log-probabilities by geometric mean. On the nine-species benchmark the held-out yeast split reaches Recall@1/5/100 of 60.0/78.9/91.9%, with neural re-scoring lifting Recall@1 to 82.9%; the harder S. brodae proteome reaches only 55.5% Recall@100, pointing to a need for larger-scale training.
- TIPs (2026): Deep-learning-guided proteogenomic framework that uses de novo sequencing to refine the search space for transposable-element-derived HLA antigens, finding 20-fold more than conventional searches.
- FastaLake (2026): Metaproteomics workflow that uses full-length AlphaNovo de novo peptide predictions as exact queries to select compact search databases from large sequence reservoirs, then searches them with Sage. Matched metagenome-derived databases in two stool cohorts and found 20-35% more peptides in intestinal fluid, without matched metagenomic sequencing.
Applied in
Neuropeptidomics, Metaproteomics, Toxin identification (biodefense), Immunopeptidomics / neoantigen
Papers describing them (11)
- Robust accurate identification of peptides (RAId): deciphering MS2 data using a structured library search with de novo based statistics (2005, Bioinformatics, peer-reviewed)
- A Hybrid, de Novo Based, Genome-Wide Database Search Approach Applied to the Sea Urchin Neuropeptidome (2010, Journal of Proteome Research, peer-reviewed)
- Identification of phosphopeptides with unknown cleavage specificity by a de novo sequencing assisted database search strategy (2014, PROTEOMICS, peer-reviewed)
- MetaNovo: a probabilistic approach to peptide discovery in complex metaproteomic datasets (2019, bioRxiv, preprint)
- Identification of the Cryptic HLA-I Immunopeptidome (2020, Cancer Immunology Research, peer-reviewed)
- MetaNovo: An open-source pipeline for probabilistic peptide discovery in complex metaproteomic datasets (2023, PLOS Computational Biology, peer-reviewed)
- Orthrus: an AI-powered, cloud-ready, and open-source hybrid approach for metaproteomics (2024, bioRxiv, preprint)
- Identification of Unknown Biological Toxin Proteins Using Mass Spectrometry: A Case Study on De Novo Sequencing of Ricin (2025, Toxins, peer-reviewed)
- INSearch: AI-native framework for large scale proteomic database search via contrastive joint embeddings and transformer-based scoring function (2026, thesis)
- TIPs: a deep learning-guided proteogenomic framework to expand the landscape of transposable element-derived antigens with immunopeptidomics (2026, Genome Biology, peer-reviewed)
- FastaLake selects compact metaproteomic databases from large sequence reservoirs by de novo peptide lookup (2026, bioRxiv, preprint)