The myNEO technology

The myNEO differential screening platform compares tumour and healthy cells (or any 2 cell types by extension) to identify immunogenic alterations present in a specific patient. These are alterations that result in tumour-unique surface peptides, absent from all healthy cells of the patient, that can trigger an immunogenic reaction against the tumour cells. Therapies using these tumour targets (neoantigens) can induce broad, specific, durable immune responses, leading to tumour regression. The myNEO bioinformatic analysis is largely based on patient-specific sequencing datasets, potentially supplemented with extra assay data for improved sensitivity and specificity. More information can be read on the technical wireframe of the neoantigen prediction platform and its variety of advanced features.

"The tools developed by myNEO allow extensive exploration of the tumour antigen landscape, while machine-learning based predictions enable high-quality selection of the optimal targets."

Identifying per patient which alterations lay at the origin of the cancer cells and which have been accumulated during clonal expansion of the tumour is of high importance. Dissimilarities between tumour and normal cells are explored on multiple levels: genomic differences caused by mutations on DNA level, transcriptomic changes due to RNA alterations, and proteomic changes due to alternative ribosomal binding or proteasomal processing are some prime examples.  In doing so, the platform discovers an as broad set of surface antigens as possible for every patient, including several novel antigen types (RTEs, gene fusions, alternative ORFs, etc.). By exploring such a broad antigen landscape, the platform finds more valuable targets, even in patients with a cold lowly mutated tumour where standard immunotherapies deliver no benefit.  

To accommodate rapid validation of the predicted antigens, myNEO has developed neoMS and neoIM. These two machine learning algorithms highlight the synergy possible between analytical and computational tools, and allow rapid selection of the optimal targets, without the need for extensive in-vitro lab tests per patient. The neoMS algorithm is a cross-patient employable deep-learning algorithm trained on over 2.5 million datapoints, that predicts the probability that the antigen will be presented (in sufficient amounts) on the surface of the tumour cells. The complementary neoIM algorithm predicts the immunogenicity of these surface peptides across populations.

Furthermore, the ImmunoEngine analyses the tumour microenvironment and its immune signature, contributing to a deeper understanding of the tumour phenotype. A differential expression study is performed on an extensive set of relevant genes as well.

The technology platform is deeply integrated with clinical benefit. After each analysis, the efficacy of specific targets is reported, continuously enhancing the platform in its consequent predictions.

The bioinformatic analysis is integrated within an end-to-end solution focused on delivering high-quality, consistent datasets within clinical timeframes. It ranges from sample dissection to construct design, wherein tumour antigens are grouped in constructs by evaluating their potential to induce both CD4 and CD8 immune responses. The personalised mRNA manufacturing technology allows for rapid epitope production in low quantities, while providing a compound capable of inducing broad and durable immune responses.

The results of the analysis are visualised in an interactive html report, to enable manual inspection of the predicted targets, and streamline scientific discussions.