Summary
A bioRxiv preprint describes NeoOmni Atlas, a database built by reanalysing mass-spectrometry data from 26 cancer types. It catalogues HLA-presented peptides and computationally prioritises potential tumour-associated and tumour-specific targets.
A bioRxiv preprint introduces NeoOmni Atlas, a database of peptides detected on human leukocyte antigen (HLA) molecules across 26 cancer types. The authors built it by reanalysing 7,561 immunopeptidomics datasets, comprising 511 million raw spectra from 2,341 samples. Their analysis catalogues peptides from conventional and non-canonical sources, then uses computational filtering to identify potential tumour-associated and tumour-specific epitopes.
Neoantigens are peptide fragments that can arise from changes in a tumour’s molecular makeup. When displayed by HLA molecules on a cell’s surface, some can be recognised by T cells, making them candidates for personalised cancer immunotherapy. A database of peptides detected in this way can help researchers identify and compare possible targets; the detection itself does not establish that a patient’s immune system will respond to them.
A larger catalogue of HLA-presented peptides
Immunopeptidomics analyses the peptides bound to HLA molecules, often using mass spectrometry to identify them. HLA proteins display peptide fragments at the cell surface, where immune cells can encounter them. The resource combines this mass-spectrometry-supported evidence with proteogenomic analysis, which links peptide findings to molecular information about their possible origins.
The authors report 872,414 eluted canonical peptides and 198,961 non-canonical peptides, spanning 18 molecular alteration classes. Across these results, the database contains 759,386 unique ligands associated with 322 HLA alleles. The authors say that, compared with existing repositories, the atlas covers more cancer types, includes 60 additional HLA alleles and adds 369,781 novel ligands.
These counts describe different parts of the catalogue: peptide detections, unique HLA ligands and the range of molecular sources represented. The breadth may help researchers search for candidate targets across cancer types and HLA backgrounds, which matter because peptide presentation depends in part on a person’s HLA molecules.
Prioritising possible targets
A computational filtering framework classified 316,126 epitopes as tumour-associated and 61,231 as tumour-specific. The authors also report 65 conserved tumour-specific epitopes recurring across more than 10 tumour types, which they propose as potential shared-antigen candidates. The atlas prioritises targets including PMEL, represented by the peptide gp100:209-217, and CSPG4.
The preprint reports stronger predicted HLA-binding affinity for tumour-specific than tumour-associated epitopes for HLA class I A and B alleles. Binding prediction is one way to rank candidates; it is not a measurement of T-cell activity. Likewise, a peptide detected in a sample is evidence of presentation in that dataset, not by itself evidence that targeting it will produce a useful immune response. The work is a database and computational analysis, rather than a clinical study of an immunotherapy.
NeoOmni Atlas is freely accessible and supports browsing, searching and downloading by cancer type, HLA allele, peptide sequence and source event. Its catalogue provides a starting point for selecting candidates for further study, while the therapeutic relevance of individual entries depends on subsequent biological and clinical evidence.