In oandrefonseca/crossdome: An R package to measure cross-reactivity risk in the sequence-space

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/",
  out.width = "100%"
)

options(tibble.print_min = 5, tibble.print_max = 5)

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.

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)

View(result@result)

library(knitr)
library(kableExtra)
library(dplyr)

to_display_columns <- c('rank', 'query', 'subject', 'n_positive',   'n_mismatch',   'relatedness_score', 'pvalue', 'hla_allele')
to_display_data <- result@result[1:30, to_display_columns] %>%
  mutate(
    relatedness_score = round(relatedness_score, 2),
    hla_allele = sub('\\*', '', hla_allele)
  )

kable(
  to_display_data, 
  row.names = FALSE
)