OmniNeo: a multi-omics pipeline incorporating proteomics and AI selection for neoantigen optimization in tumor immunotherapy.

Neoantigen-based vaccines represent a promising approach in cancer immunotherapy, with the key to their effective clinical application lying in the precise identification of immunogenic neoantigens. Existing methods primarily focus on genomic variations, lacking integration of multi-omics data and essential filtering steps, which limits comprehensive assessment of immunogenicity and results in only a small subset of neoantigens capable of eliciting effective immune responses. Moreover, the complexity and poor portability further hinder the clinical applicability. To address these limitations, we developed OmniNeo, an automated multi-omics-based neoantigen discovery framework. 1) OmniNeo integrates whole-genome/exome sequencing (WGS/WES), transcriptomic, and proteomics data to simultaneously identify neoantigenic epitopes derived from SNVs/Indels, frameshift mutations, gene fusions, and non-coding region variations; 2) The pipeline incorporates a convolutional neural network-based model, OmniNeo-CNN along with multiple filtering mechanisms to quantify the immunogenicity and T-cell receptor (TCR) recognition potential of predicted neoantigen candidates through multiple features; 3) The workflow is built on nextflow, offering a one-stop, scalable, and portable solution for rapid and efficient neoantigen prediction. Finally, we demonstrated the practical application procedures of this workflow in potential tumor immunotherapy through case study analyses of liver cancer samples. The tool is freely accessible as an open-source resource via https://github.com/linfengxu/OmniNeo, https://zenodo.org/records/15340824.