Calculate neoantigen summary statistics for priority, classic, alternative, frameshift-derived, and fusion-derived neoantigens for each sample.




Data table. Prediction output from garnish_affinity.


Summary table description:

A summary data table of dt by sample_id with the following columns:

  • classic_neos: classically defined neoantigens (CDNs); defined as mutant nmers predicted to bind MHC with high affinity (< 50nM \(IC_{50}\))

  • classic_top_score: sum of the top three mutant nmer affinity scores (see below)

  • alt_neos: alternatively defined neoantigens (ADNs); defined as mutant nmers predicted to bind MHC with greatly improved affinity relative to non-mutated counterparts (10x for MHC class I and 4x for MHC class II) (see below)

  • alt_top_score: sum of the top three mutant nmer differential agretopicity indices; differential agretopicity index (DAI) is the ratio of MHC binding affinity between mutant and wt peptide (see below). If the peptide is fusion- or frameshift-derived, the binding affinity of the closest wt peptide determined by BLAST is used to calculate DAI.

  • fs_neos: mutant nmers derived from frameshift variants predicted to bind MHC with < 500nM \(IC_{50}\)

  • fusion_neos: mutant nmers derived from fusion variants predicted to bind MHC with < 500nM \(IC_{50}\)

  • IEDB_high: defined as mutant nmers predicted to bind MHC with affinity (< 500nM \(IC_{50}\)) and IEDB alignment score > 0.9.

  • high_dissimilarity: defined as mutant nmers predicted to bind MHC with affinity (< 500nM \(IC_{50}\)) and dissimilarity > 0.75.

  • nmers: total mutant nmers created

  • predictions: wt and mutant predictions performed

  • mhc_binders: nmers predicted to at least minimally bind MHC (< 500nM \(IC_{50}\))

  • variants: total genomic variants evaluated

  • transcripts: total transcripts evaluated

Additional information

Differential agretopicity index (DAI) expresses the degree to which peptide binding to MHC class I or II differ due to the presence of a non-synonymous mutation. Alternatively defined neoantigens (ADNs) are mutant peptides predicted to bind MHC class I or II with greatly improved affinity relative to non-mutated counterparts (i.e. peptides with high DAI). In mice, selection of peptides with high DAI results in a substantially improved rate of experimentally validated epitopes that mediate protection from tumor growth.

To determine DAI, mutant peptides that at least minimally bind MHC class I or II (> 500nM \(IC_{50}\)) are selected and then DAI is calculated as the fold-change in binding affinity between non-mutant and mutant peptides. ADNs are identified as mutant peptides with DAI > 10 for MHC class I and > 4 for class II.

ADNs are generated from selective mutations in the peptide-MHC anchor position (i.e. the agretope) rather than mutations randomly occurring across the peptide sequence. This feature leads to two potentially important and unique immunological characteristics of ADNs. First, unlike CDNs, the TCR-facing peptide sequence in ADNs is likely the same as the corresponding non-mutant peptide. Second, the MHC binding of the corresponding non-mutant peptide may be so low that its presentation in the thymus is minimal and central tolerance may be bypassed. Functionally, there is evidence of strong immunogenicity of ADNs. Peptides, which in retrospect satisfy ADN selection criteria, are common among human tumor antigens that have been experimentally confirmed to be immunogenic. A recent extensive analysis of tumor immunity in a patient with ovarian carcinoma showed that the top five reactive mutant peptides had substantially higher mutant to non-mutant predicted MHC class I binding affinity. Moreover, a re-analysis of validated neoantigens from non-small cell lung carcinoma or melanoma patients showed that one third of these were ADNs (resulting from an anchor position substitution that improved MHC affinity > 10-fold).

To better model potential for oligoclonal antitumor responses directed against neoantigens, we additionally report a top three neoantigen score, which is defined as the sum of the top three affinity scores \(\left(\frac{1}{IC_{50}}\right)\) for CDNs or sum of top three DAI for ADNs. The top three was chosen in each case because this is the minimum number that captures the potential for an oligoclonal T cell response and mirrors experimentally confirmed oligoclonality of T cell responses against human tumors. Moreover, the top three score was the least correlated to total neoantigen load (vs. top 4 through top 15) in a large scale human analysis of neoantigen across 27 disease types (R-squared = 0.0495), and therefore not purely a derivative of total neoantigen load.


Carreno BM, Magrini V, Becker-Hapak M, Kaabinejadian S, Hundal J, et al. 2015. A dendritic cell vaccine increases the breadth and diversity of melanoma neoantigen-specific T cells. Science. 348(6236):aaa3828–808

Duan F, Duitama J, Seesi Al S, Ayres CM, Corcelli SA, et al. 2014. Genomic and bioinformatic profiling of mutational neoepitopes reveals new rules to predict anticancer immunogenicity. Journal of Experimental Medicine. 211(11):2231–48

Fritsch EF, Rajasagi M, Ott PA, Brusic V, Hacohen N, Wu CJ. 2014. HLA-binding properties of tumor neoepitopes in humans. Cancer Immunology Research. 2(6):522–29

Gros A, Parkhurst MR, Tran E, Pasetto A, Robbins PF, et al. 2016. Prospective identification of neoantigen-specific lymphocytes in the peripheral blood of melanoma patients. Nature Medicine. 22(4):433–38

Jiménez-Sánchez A, Memon D, Pourpe S, Veeraraghavan H, Li Y, et al. 2017. Heterogeneous Tumor-Immune Microenvironments among Differentially Growing Metastases in an Ovarian Cancer Patient. Cell. 170(5):927–938.e20

McGranahan N, Furness AJS, Rosenthal R, Ramskov S, Lyngaa R, et al. 2016. Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science. 351(6280):1463–69

Verdegaal EME, de Miranda NFCC, Visser M, Harryvan T, van Buuren MM, et al. 2016. Neoantigen landscape dynamics during human melanoma-T cell interactions. Nature. 536(7614):91–95

See also


if (FALSE) { library(magrittr) library(data.table) library(antigen.garnish) # load an example VCF dir <- system.file(package = "antigen.garnish") %>% file.path(., "extdata/testdata") dt <- "antigen.garnish_example.vcf" %>% file.path(dir, .) %>% # extract variants garnish_variants %>% # add space separated MHC types # see list_mhc() for nomenclature of supported alleles .[, MHC := c("HLA-A*01:47 HLA-A*02:01 HLA-DRB1*14:67")] %>% # predict neoantigens garnish_affinity # summarize predictions dt %>% garnish_summary %T>% print # generate summary graphs dt %>% garnish_plot }