Back

DelPi Learns Generalizable Peptide-Signal Correspondence for Mass Spectrometry-Based Proteomics

Park, J.; Kim, K.; Kang, U.-B.; Kim, S.

2026-01-07 bioinformatics
10.64898/2026.01.06.697814 bioRxiv
Show abstract

Peptide identification in mass spectrometry-based proteomics has traditionally relied on handcrafted features or simplified probabilistic approaches that limit the interpretation of structured peptide evidence. We present DelPi, an open-source peptide identification framework that learns generalizable peptide-signal correspondence from raw spectra through self-supervised pre-training followed by task-specific fine-tuning. With model distillation enabling practical deployment, DelPi expands the interpretation of peptide evidence across data-independent and data-dependent acquisition while maintaining robust false discovery control.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.