Neoantigen Discovery

We use machine learning and neural networks to provide you with advanced analytics to better inform your discovery and translational research programs.

Leveraging our enhanced exome and transcriptome, we’ve applied these assays for deeper bioinformatics analysis to accurately predict neoantigens.  Using neoantigens as a predictor for determining likelihood of response to immunotherapies is one exciting application (Rizvi et al., 2015), but also using these analyses to develop personalized cancer vaccines is increasingly critical.  Although standard research pipelines have been developed to assist in neoantigen identification, developing a strong analytically validated neoantigen identification platform suitable for clinical applications is complex.

Within our neoantigen workflow (Figure 4), somatic DNA and RNA variants are processed for antigen identification, including SNVs, indels, and fusion events. Importantly, both in-frame and out-of-frame events are accurately considered by transcript, allowing for a rich source of putative candidate neoantigens. Our pipeline includes assessment of potentially important criteria including HLA prediction, MHC binding (class I and II), immunogenicity, similarity to self, and similarity to known antigens.

Further, peptides are evaluated for both gene and variant-level expression.

Taken together, the ACE ImmunoID product provides a comprehensive assessment of features allowing the ability to identify and rank potentially immunogenic neoantigens.

Figure 4

The NeoantigenID enables the identification of neoantigens and allows you to assess:

  • The mutational landscape with neoantigen load and mutational burden
  • The peptides derived from SNVs, indels and gene fusions
  • Critical metrics such as:
    • Gene- and variant-level expression
    • MHC Class I and select Class II binding affinity
    • Immunogenicity

Correct HLA Typing for accurate neoantigen prediction

Additional features within NeoantigenID are the in silico HLA Typing of both Class I and select Class II alleles and further refinement of peptide prediction by the incorporation of phasing within the pipeline.

HLA typing is an essential component of the neoantigen prediction process. Incorrect typing can lead to downstream inaccuracies in binding predictions. In order to provide accurate HLA typing within ACE ImmunoID, we have optimized and validated an extremely accurate tool.  To accomplish, we validated hundreds of HLA alleles across both Class I and select Class II that had previously been typed with clinical grade PCR-based methods. We performed a blinded clinical validation with 62 orthogonally typed samples involving ~700 HLA alleles, achieving high concordance with the ACE Exome.

Table 1: Results of HLA Typing

HLA Locus

Number of Calls

Number in Agreement

Concordance

Class I

318

311

98%

Class II

378

355

94%

Class I + II

696

666

96%

2018-02-05T17:26:18+00:00