Reanalysing immunopeptidomics data against noncanonical sequences

Alithea Bio's case study of 11 May 2026 credits Alisa Fuchs and me with a dark-proteome analysis using HLA-Compass AI. The work asked which peptides from an expanded catalogue of noncanonical open reading frames could be found by reanalysing existing immunopeptidomics samples.

The company reports searching a catalogue of 28,359 sequences and obtaining results in seven hours. It reports 17,082 unique noncanonical peptides, including 11,553 that were not observed in the healthy-tissue samples represented in its database.

Those observations define a candidate-discovery result. Absence from the sampled healthy tissues does not establish absence across every healthy tissue or condition. Peptide detection also leaves questions about reproducibility, biological relevance, tumour selectivity and experimental validation before a therapeutic application can be assessed.

The public article is a company case study that credits our analysis. It provides a record of applying a new sequence catalogue to an existing research collection and identifying candidates for further investigation.

Read the original case study.