In Bozorgui/imogene: : Immuno-oncology gene expression associations

knitr::opts_chunk$set(collapse = TRUE,comment = "#>")
knitr::opts_chunk$set(fig.width = 4, fig.height = 4) 
options(tibble.print_min = 4L, tibble.print_max = 4L)

library(imogimap)

imogimap (Immuno-Oncology gene interaction maps)

Introduction

imogimap calculates statistical synergy scores based on gene expression profiles in multi-sample data to quantify combinatorial effects of oncogenes and immune checkpoints on immune related phenotype.

Example use of TCGA data

imogimap uses curatedTCGAData API functions to access RNASeq2GeneNorm expressions.

As an example, let us explore TGFb cytokines combinatorial interactions with immune checkpoints and their effects on all immune pheotypes/features as listed in TCGA_immune_features_list. First we define gene lists using Hugo IDs. a pre-defined list of immune checkpoints, icp_gene_list, is included as default for immune checkpoints:

my_onco <- c("TGFBR1")
my_icp <- icp_gene_list

Then we define disease lists using any of the TCGA abbreviated disease names as listed in TCGA_disease_list:

my_cohort <- c("luad")

Now we use im_syng_tcga function to get synergy scores. The rest of parameters are optional, with defualt values as below:

my_syng_df <-  im_syng_tcga(
  onco_gene  = my_onco,
  icp_gene = my_icp, 
  cohort = my_cohort, 
  method = "independence",
  sensitivity = F,specificity = F)

We have the option to choose one of the two available synergy scoring methods and determine whether we want to include specificity and robustness analysis. Specificity and robustness measures are calculated by random bootstrapping, and we can increase or decrease the number of bootstrapping iterations, N_iteration, to gain more accuracy or save time.

Other optional inputs of im_syng_tcga are: sample_list that allow us to define a curated set of TCGA samples, and feature that allow us to provide additional immune phenotypes/features.

The output contains synergy scores for all pairs of genes, as well as specificity p.values and robustness variances that are used to identify noteworthy interactions on all the immune features listed in TCGA_immune_features_list.

synergy outputs can be visualized as an igraph graphical network model for each specific phenotype using im_netplot function and saved as an image:

jpeg("syng_network.jpeg",width=1000,height=1000)
im_netplot(df = my_syng_df ,
           Immune_phenotype = "IFNGscore",
           cutoff = 0.35,
           cohort = "luad" ,
           icp_gene = my_icp,
           seed = 123)
dev.off()

im_netplot function directly uses the output of im_syng_tcga function.A unique immune feature from the output needs to be specified. We have the option to specify a sub-group of our disease cohorts, or make a single network for all the cohorts in the output. We are also able to change the cutoff value for synergy scores. Synergy scores with their absolute value below the cut-off won't be displayed. We may want to input our immune checkpoint list if it differs from imogimap's default icp_gene_list. im_netplot uses this list to color code onco-genes and immune checkpoints differently. Finally by changing seed we can change the plots layout and make it reproducible.

imogimap also visualizes data that has been used in calculation of an individual synergy scores. To do this we can use im_boxplot_tcga function:

im_boxplot_tcga(onco_gene = "TGFB1", icp_gene = "CD27",cohort = "luad",
Immune_phenotype = "IFNGscore")

im_boxplot_tcga, accepts single and unique values for each of the input parameters, with onco_gene and icp_gene being gene's Hugo ID,and cohort any of the TCGa abbreviated disease names, and Immune_Feature any of the features in TCGA_immune_features_list. The result is a boxplot of immune feature values, stratified based on gene pair expression values.

Bozorgui/imogene documentation built on Feb. 21, 2022, 11:57 a.m.