Crossdome (Beta)

Developed by: André Fonseca, PhD

Installation

devtools::install_github("antuneslab/crossdome", build_vignettes = TRUE)

Abstract

Crossdome: An interactive R package to predict cross-reactivity risk using immunopeptidomics databases

Currently, several clinical protocols are leveraging on distinct immune mechanisms, such as adoptive T-cell therapy and peptide-based vaccines. However, multiple factors can impact the accuracy of these immune-based applications, such as expression heterogeneity, immunogenicity, and cross-reactivity (CR) risk. Crossdome was created to measure cross-reactivity potential based on biochemical properties. Our approach aims to rank potential CR candidates and measure cross-reactivity risk using mRNA expression, immunogenicity score (TCR binding), and MHC presentation probability. Additionally, we provide the expression profile related to each CR candidate.

Figure 1. Crossdome workflow and strategy. Crossdome summarises biochemical properties per amino acid into 12 principal components. In turn, the principal components are used to convert peptide sequences into biochemical profiles (matrices). Next, given a target peptide, Crossdome screens an immunopeptidomics dataset for a similar biochemical profiler, i.e., CR candidates. A relatedness score between the target and candidate off-targets is calculated based on weighted linear distance. Finally, Crossdome incorporates expression levels and immunogenicity predictions for each potential off-target.

Functions and Features

Prediction

• cross_background Peptide database spanning eluted candidates (experimentally validated) and custom (user-defined).
• cross_pair_summary Calculates relatedness score between peptides.
• cross_compose Predicts relatedness among peptides in a given database. Low values are associated with cross-reactive candidates.
• cross_browser Opens an interactive shiny application.

Analysis

• cross_expression_matrix Extracts gene donor mRNA expression based on CR candidates.
• cross_substitution_matrix Calculates position-specific substitution across cross-reactive candidates.
• cross_peptide_properties Converts a peptide to biochemical profile.
• cross_write Exports Crossdome result slot for a tsv file.

Visualization

• cross_expression_plot A heatmap presenting the gene donor expression profile.
• cross_tissues_plot A bar plot summarizing the tissue-specificy groups.
• cross_prediction_plot Plot a dot plot showing the immunogenic predictions [UNDER CONSTRUCTION].
• cross_pairwise_plot Correlation plot based on two peptide / biochemical profiles [UNDER CONSTRUCTION].
• cross_substitution_plot A heatmap combined with seqlogo displaying amino acid substitutions.

Documentation

Examples of each function, including code, are available in the package documentation. Please use: ?FUNCTION_NAME. Additionally, two vignettes are available through the command browseVignettes('crossdome').

Included datasets

• hla_database Immunopeptidomics spanning several MHC Class I alleles. Includes binding affinity, and immunogenicity score from MHC Flurry and DeepImmuno, respectively.
• hpa_database Expression database derived from Human Protein Atlas.
• peptide_annotation Database for mapping peptides to gene-donors derived from NCBI RefSeq Protein [UNDER CONSTRUCTION].
• mage_off_targets Curated off-targets related to MAGEA3-specific TCR.

To inspect the data sets use: data(DATA_NAME)

Basic Usage

library(crossdome)

database <- cross_background(off_targets = 'ESDPIVAQY', allele = "HLA-A*01:01")
result <- cross_compose(query = 'EVDPIGHLY', background = database)
#> ##------ Mon Mar 27 12:10:59 2023 ------##

View(result@result)

rank	query	subject	n_positive	n_mismatch	relatedness_score	pvalue	hla_allele
1	EVDPIGHLY	EVDPIGHLY	9	0	0.00	0.0000000	HLA-A01:01
2	EVDPIGHLY	EVDPIGHVY	9	1	1.35	0.0000001	HLA-A01:01
3	EVDPIGHLY	EADPTGHSY	6	3	8.72	0.0000336	HLA-A01:01
4	EVDPIGHLY	EVDPTSHSY	6	3	10.43	0.0001107	HLA-A01:01
5	EVDPIGHLY	VSDPVGVLY	6	4	10.95	0.0001567	HLA-A01:01
6	EVDPIGHLY	EMDPVTQLY	8	4	11.23	0.0001889	HLA-A01:01
7	EVDPIGHLY	YTDPVGVLY	6	4	11.53	0.0002293	HLA-A01:01
8	EVDPIGHLY	EHDPVGQMV	7	5	11.69	0.0002542	HLA-A01:01
9	EVDPIGHLY	ESDPEGALW	6	4	12.72	0.0004821	HLA-A01:01
10	EVDPIGHLY	EVAPAGASY	5	4	12.95	0.0005563	HLA-A01:01
11	EVDPIGHLY	NVDPVQHTY	6	4	12.97	0.0005612	HLA-A01:01
12	EVDPIGHLY	DSDPTGTAY	5	5	13.04	0.0005853	HLA-A01:01
13	EVDPIGHLY	DVSPLMHLF	7	5	13.49	0.0007682	HLA-A01:01
14	EVDPIGHLY	QIDPVLQVY	8	6	13.55	0.0007944	HLA-A01:01
15	EVDPIGHLY	SLDPLAMLY	6	5	13.58	0.0008085	HLA-A01:01
16	EVDPIGHLY	SVDPHGFIS	5	5	13.60	0.0008169	HLA-A01:01
17	EVDPIGHLY	LLDPIKELY	6	4	13.66	0.0008489	HLA-A01:01
18	EVDPIGHLY	FVEPVGLNY	6	5	13.75	0.0008957	HLA-A01:01
19	EVDPIGHLY	FVNPISFVY	6	5	13.88	0.0009614	HLA-A01:01
20	EVDPIGHLY	KFDPVGEIL	6	6	14.04	0.0010592	HLA-A01:01
21	EVDPIGHLY	AVDPIKDMY	6	4	14.06	0.0010678	HLA-A01:01
22	EVDPIGHLY	FLEPLGLAY	6	6	14.08	0.0010839	HLA-A01:01
23	EVDPIGHLY	QLDPIYVAY	6	5	14.15	0.0011284	HLA-A01:01
24	EVDPIGHLY	ELDPNNAVY	6	5	14.18	0.0011473	HLA-A01:01
25	EVDPIGHLY	YVDSEGHLY	6	3	14.20	0.0011600	HLA-A01:01
26	EVDPIGHLY	IISPVGDAY	5	6	14.37	0.0012774	HLA-A01:01
27	EVDPIGHLY	ESDPIVAQY	5	4	14.38	0.0012840	HLA-A01:01
28	EVDPIGHLY	NIDPITMAY	5	5	14.38	0.0012849	HLA-A01:01
29	EVDPIGHLY	QVDPLSALK	6	5	14.46	0.0013432	HLA-A01:01
30	EVDPIGHLY	KWDPTGHIL	6	5	14.47	0.0013482	HLA-A01:01

str(result)
#> Formal class 'xrResult' [package "crossdome"] with 7 slots
#>   ..@ query          : chr "EVDPIGHLY"
#>   ..@ result         :'data.frame':  37656 obs. of  10 variables:
#>   .. ..$ rank             : int [1:37656] 1 2 3 4 5 6 7 8 9 10 ...
#>   .. ..$ query            : chr [1:37656] "EVDPIGHLY" "EVDPIGHLY" "EVDPIGHLY" "EVDPIGHLY" ...
#>   .. ..$ subject          : chr [1:37656] "EVDPIGHLY" "EVDPIGHVY" "EADPTGHSY" "EVDPTSHSY" ...
#>   .. ..$ n_positive       : int [1:37656] 9 9 6 6 6 8 6 7 6 5 ...
#>   .. ..$ n_mismatch       : int [1:37656] 0 1 3 3 4 4 4 5 4 4 ...
#>   .. ..$ relatedness_score: num [1:37656] 0 1.35 8.72 10.43 10.95 ...
#>   .. ..$ zscore           : num [1:37656] -5.48 -5.25 -3.99 -3.69 -3.6 ...
#>   .. ..$ pvalue           : num [1:37656] 2.12e-08 7.66e-08 3.36e-05 1.11e-04 1.57e-04 ...
#>   .. ..$ hla_allele       : chr [1:37656] "HLA-A*01:01" "HLA-A*01:01" "HLA-A*01:01" "HLA-A*01:01" ...
#>   .. ..$ percentile_rank  : num [1:37656] 0 0.00266 0.00531 0.00797 0.01062 ...
#>   ..@ allele         : chr "HLA-A*01:01"
#>   ..@ expression     : list()
#>   ..@ analysis       : list()
#>   ..@ position_weight: num [1:9] 1 1 1 1 1 1 1 1 1
#>   ..@ timestamp      : chr "##------ Mon Mar 27 12:10:59 2023 ------##"