Back

Integrated immunopeptidomics and transcriptomics reveal a cancer-restricted non-canonical HLA class I ligandome in medulloblastoma

Hui, A.; Zheng, N.; Xu, Y.; Mai, W.

2026-09-28 cancer biology
10.64898/2026.09.26.754718 bioRxiv
Show abstract

Medulloblastoma is the leading type of malignant brain tumor in children. Despite recent advances in T cell-based immunotherapy, the low mutational burden of medulloblastoma poses a considerable challenge to the discovery of mutation-derived neoantigens. Previous studies of medulloblastoma have investigated tumor-associated antigens and neoantigens derived from aberrant splice junctions. The landscape of non-canonical HLA class I-bound peptides (ncHLAp) remains poorly defined. To broaden the repertoire of targetable antigens in medulloblastoma, we mapped the ncHLAp landscape across the four major molecular subgroups of medulloblastoma (Group 3, Group 4, WNT, and SHH) by integrating immunopeptidomic and transcriptomic analyses. Candidates were further filtered against normal-tissue immunopeptidomic and ribosome profiling datasets to identify cancer-restricted ncHLAp (CR ncHLAp). A total of 2,298 ncHLAp were identified across 21 human medulloblastoma samples, of which 598 were CR ncHLAp. These CR ncHLAp were primarily derived from non-canonical translational sources, including retained introns, out-of-frame open reading frames, and long non-coding RNAs. Some CR ncHLAp were shared across medulloblastoma samples and detected in other cancer types. More than 70% of the CR ncHLAp were predicted to bind at least one patient-matched HLA allotype, and approximately 87% were predicted to have immunogenic potential. Collectively, this study defines a previously undercharacterized component, CR ncHLAp, of the medulloblastoma HLA-I ligandome and suggests that CR ncHLAp may serve as promising candidate targets for T-cell-based immunotherapy in medulloblastoma.

Matching journals

The top 6 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.