Predicting and identifying neoantigens involves a multidisciplinary approach encompassing genomics, bioinformatics, and immunology.

Upon diagnosis of the patient, the process initiates with the selection of tumor tissue samples. These samples are meticulously curated to represent areas abundantly populated by tumor cells. Histological evaluation is of paramount importance in distinguishing cancerous tissue from healthy tissue and offers valuable insights into the tumor’s histopathological attributes, encompassing its type, grade, and stage. Once histological analysis definitively confirms the presence of tumor cells, the selection of tissue becomes pivotal. Careful consideration is given to identifying tumor-rich regions for subsequent genomic and proteomic analyses. These regions are expected to harbor the specific genetic mutations and neoantigens pertinent to the patient’s cancer, rendering them the focal point for personalized immunotherapies.

The tumor is biopsied, and both the cancerous and normal tissues are sequenced. Whole exome sequencing (WES), which is a powerful genomic technique, focuses on sequencing the coding regions, or exons, of genes in an individual’s genome. While the majority of the human genome consists of non-coding regions, exons are where most disease-associated mutations are found. By sequencing only the exonic regions, WES enables researchers to identify single nucleotide variations, small insertions, and deletions [27,28]. Unlike WES, which focuses on the genome’s DNA, transcriptome analyses, or RNA sequencing (RNA-seq) examines the transcripts that are produced from genes and serve as templates for protein synthesis. Specifically, RNA-Seq can detect alternative splicing events, where different exons are included or excluded from the final mRNA transcript and result in the production of diverse protein isoforms. Additionally, RNA-Seq captures gene fusions arising from chromosomal rearrangements, which lead to the expression of fusion transcripts that are translated into fusion proteins with unique antigenic properties. RNA-seq involves extracting RNA molecules from cells or tissues and converting them into complementary DNA (cDNA) through reverse transcription. These cDNA fragments are then sequenced to quantify the abundance of various RNA molecules [29]. The incorporation of RNA-Seq into mutanome analyses enables prioritization of highly expressed mutations over nonexpressed variants, resulting in a more refined selection of potential vaccine candidates.

Mutation calling is a computational process that follows the sequencing of DNA (WES) or cDNA (RNA-seq) to identify genetic variations or mutations within the genome. The goal is to distinguish between the naturally occurring variations and mutations that might contribute to disease. The process involves several steps, including aligning sequenced reads to a reference genome, identifying areas with discrepancies (variations), and filtering out background noise and false positives. Currently, specialized algorithms like Varscan, SomaticSniper, Strelka, and MuTect2 GATK are employed to identify somatic mutations by comparing tumor and normal sequences. They identify somatic mutations that are unique to the tumor and not present in the individual’s normal tissue. Filters are applied to remove common germline variations and retain tumor-specific alterations. By comparing the tumor genome (WES) or transcriptome (RNA-seq) with the corresponding normal tissue, researchers can pinpoint mutations that have arisen during tumorigenesis [30].

Following mutation calling, the subsequent annotation steps are crucial for neoantigen prediction. Annotation involves assigning functional and contextual information to the identified variants using tools like ANNOVAR, Variant Effect Predictor (VEP), or SnpEff. These tools help determine whether a variant falls within a protein-coding region, its effect on the amino acid sequence, and its potential impact on protein structure and function. Subsequently, the annotated variants are further analyzed to identify mutations that generate altered peptide sequences (neoepitopes) capable of binding to MHC molecules and eliciting an immune response. The integration of mutation calling and comprehensive variant annotation lay the foundation for accurately predicting potential neoantigens that can be harnessed for personalized cancer immunotherapy strategies [31,32].

Epitope prediction algorithms play a pivotal role in neoantigen prediction by evaluating the likelihood that a given peptide sequence, derived from a mutated protein, will bind to MHC molecules with sufficient affinity to be presented to T cells. Currently, multiple tools such as SYFPEITHI, IEDB, and NetMHCpan are extremely useful in offering unique features in epitope-binding prediction. For instance, SYFPEITHI calculates the binding affinity of peptides to MHC class I molecules. It employs a scoring system based on experimental binding data to predict the likelihood of peptide-MHC interaction. This approach enables the identification of potential neoepitopes that have a high probability of being presented by MHC molecules and recognized by T cells. SYFPEITHI’s approach, while valuable, is often constrained by the availability of experimental binding data for a diverse range of MHC alleles [33]. While SYFPEITHI, Rankpep, and BIMAS served as pioneering prediction tools, the field has seen the emergence of more refined alternatives. Among these, NetMHC stands out as one of the most widely utilized and rigorously validated algorithms available today. NetMHC employs artificial neural networks to predict peptide binding across various MHCI variants, yielding the predicted IC50 as an output. The accuracy of neural network-based methods relies on the training set’s quality and size, thus performing better for more prevalent alleles. Notably, a refined version known as NetMHCpan expands the training dataset to encompass data from diverse species, enhancing the accuracy of predictions for less common MHC alleles [34].

Currently, the most useful epitope prediction algorithms are those focusing on peptide binding to MHC class I molecules. The MHCI antigen presentation pathway plays a central role in presenting peptides derived from endogenous cellular proteins to CD8+ T cells. Intracellular proteins undergo proteasomal processing, yielding 8–11 amino acid peptides that are subsequently transported into the endoplasmic reticulum (ER) by the transporter associated with antigen processing. There, they associate with newly synthesized class I molecules, forming stable peptide–MHCI complexes that are transported to the cell surface [35,36]. On the other hand, MHC class II antigen presentation involves the presentation of peptides derived from exogenous antigens, often proteins internalized through endocytosis or phagocytosis. In the endosomal compartments of antigen-presenting cells, these antigens are processed into peptide fragments, and a subset of these peptides binds to MHC II molecules within the groove created by the α and β chains. The resulting peptide-MHC II complex is then transported to the cell surface, where it is presented to CD4+ T helper cells. While prediction algorithms for MHCI neoantigens have flourished, MHC II neoantigens have posed challenges due to their diverse lengths (ranging from 13 to 25 amino acids) and increased binding complexity. As a result, there is a relative scarcity of binding-affinity training data and fewer algorithms available for predicting MHC II neoantigens [3].

Subsequent to neoantigen prediction using NetMHCpan, a pivotal step involves the comprehensive prioritization of neoepitopes, ensuring the selection of those with the highest potential to trigger a robust immune response. While NetMHCpan aids in identifying peptide sequences likely to bind to MHC molecules, further criteria are considered to assess their immunogenicity. Among the steps in neoepitope prioritization, the predicted binding affinity holds significance, as neoepitopes displaying strong MHC binding are more likely to be presented to immune cells. Tools like MHCflurry and MHCconsortium can also refine binding affinity predictions, aiding in the identification of top candidates. Additionally, assessing the conservation of the mutated amino acid across species using tools such as SIFT and PolyPhen enhances the understanding of its potential functional impact. Estimating epitope abundance is presently achieved through an indirect assessment involving the quantification of RNA expression levels. Mutations can be identified through tumor-to-normal DNA comparisons undergo bioinformatic scrutiny to gauge their immunogenic potential. The subsequent estimation of candidate immune stimulatory peptide levels is facilitated by RNA-Seq analysis. Prioritization based on gene expression levels, using databases like The Cancer Genome Atlas and Genotype-Tissue Expression, adds another layer of insight into neoepitope selection. Furthermore, considering the antigen processing machinery’s efficiency, tools like NetChop and NetCTLpan evaluate proteasomal cleavage and T cell processing, respectively. Neoepitopes arising from frameshift mutations and non-synonymous alterations are often prioritized due to their potential to generate immunogenic peptides. Tailoring the prioritization based on tumor heterogeneity and patient-specific HLA type refines the strategy [37].

Validation of immunogenic neoantigens can be performed in many ways, among which the most common methodologies included are mass spectrometry, tetramer/multimer staining, and ELISpot, ELISA, or intracellular cytokine staining. By eluting bound peptides and identifying using tumor-specific variant libraries, mass spectrometry is able to profile the neoantigens presented on the MHC molecules. Its high sensitivity enables the detection of even minute quantities of antigens, making it well-suited for the task. Moreover, it offers an unbiased approach, capable of identifying a wide range of neoantigens without prior knowledge of their sequences. Mass spectrometry can also provide quantitative data about the abundance of neoantigens, which is valuable for assessing their significance. However, it comes with certain complexities. Specialized equipment and expertise are prerequisites, making it less accessible for some laboratories. Additionally, sample preparation can be time-consuming and technically challenging. It is worth noting that mass spectrometry primarily detects neoantigens presented on MHC class I molecules, limiting its applicability to this subset of antigens [38].

Complementary to mass spectrometry, tetramer or multimer staining facilitates the visualization and quantification of neoantigen-specific T cells. This technique employs fluorescently labeled MHC–peptide complexes to detect and enumerate neoantigen-specific T cell populations, offering insights into their abundance and specificity and confirming the presence of T cells capable of recognizing the neoantigens. However, there are limitations to consider. Tetramer and multimer staining require prior knowledge of the neoantigens of interest and the availability of corresponding tetramers/multimers. Custom production of these reagents can be expensive and time-consuming. Moreover, these techniques may have limited sensitivity in detecting rare T cell populations, posing challenges in studies where low-frequency neoantigen-specific T cells are of interest [39].

Functional validation techniques like ELISpot, ELISA, and intracellular cytokine staining assess the ability of neoantigens to stimulate T cell responses. With synthesized neoepitopes, ELISpot and ELISA quantify interferon-gamma secretion or cytokine production in response to neoantigens, providing quantitative data on T cell activation. Meanwhile, intracellular cytokine staining detects cytokine production within T cells, corroborating their activation status. However, they do have limitations to consider. ELISpot and ELISA may lack the specificity of tetramer staining and mass spectrometry, potentially leading to false-positive results. Intracellular cytokine staining, while capable of detecting functional responses, may not directly identify neoantigen-specific T cells. Additionally, the sensitivity of these techniques may be limited, particularly in detecting low-frequency neoantigen-specific T cell populations [40].

In conclusion, the choice of neoantigen validation technique should align with the specific goals of the study, available resources, and the nature of the neoantigens under investigation. Each technique has its strengths and limitations, and combining multiple methods can provide a comprehensive assessment of neoantigen-specific immune responses, enhancing the overall validation process.