TIL clusters: immune properties

Setup

library(ithi.utils)
load_base_libs()

library(methods)
library(ComplexHeatmap)
library(cluster)
library(entropy)
library(grid)
library(gridExtra)
library(gridBase)

library(ithi.meta)
library(ithi.figures)
library(ithi.utils)
library(ithi.supp)
library(ithi.clones)
library(ithi.xcr)

ihc_table_path <- snakemake@input$ihc_table
xcr_table_path <- snakemake@input$xcr_table
molecular_subtype_file <- snakemake@input$molsubtypes
tcr_diversity_file <- snakemake@input$tcr_diversity
bcr_diversity_file <- snakemake@input$bcr_diversity

tils_for_cluster <- snakemake@params$tils_for_cluster
all_tiltypes <- snakemake@params$all_tiltypes
db_path <- snakemake@params$db

annotation_colours <- ithi.figures::get_annotation_colours()
ihc_table <- fread(ihc_table_path)
molsubtypes <- fread(molecular_subtype_file)
xcr_table <- read_clonotypes(xcr_table_path, duplicates = FALSE, db_path = db_path)

Read 16.4% of 304822 rows
Read 55.8% of 304822 rows
Read 85.3% of 304822 rows
Read 304822 rows and 18 (of 18) columns from 0.070 GB file in 00:00:06

Analysis

This reviewer makes a great point here – it would make more sense for us to do this with our subtypes. My original intention was to emphasize novelty – i.e. in showing that C1 tumours have higher TCR diversity, which is not known AFAIK in the literature – but we could just as easily do this without own subtypes.

So let’s do so.

til_subtype_results <- ithi.supp::supp_molsubtype_immune_properties(ihc_table, 
    molsubtypes, xcr_table, tcr_diversity_file, bcr_diversity_file, db_path, 
    tiltypes = all_tiltypes, subtype_class = "til_clusters")

grid.newpage()
grid.draw(til_subtype_results$fig[[1]])

grid.newpage()
grid.draw(til_subtype_results$fig[[2]])

So it seems that, although S-TIL is more diverse than ES-TIL, the difference is not significant, except in the case of BCRs when using the D50 index. What that would mean is that BCR repertoires are dominated by rare clonotypes in S-TIL samples, to a greater degree than in ES-TIL and N-TIL samples.

Nevertheless, we can see a consistent trend towards higher TCR and BCR diversity values in S-TIL and ES-TIL subsets. This is a finding that may be validated by further studies/larger cohorts.

Next, we’ll just show the TIL densities in each of the TIL subtypes. This used to show TIL densities in each of the molecular subtypes.

grid.newpage()
grid.draw(til_subtype_results$combined_fig)

As expected, epithelial TIL densities are higher in ES-TIL, and stromal TIL densities are highest in S-TIL (but for CD8+’s and plasma comparable between S-TIL and ES-TIL).

Additionally, there’s another pretty interesting finding here – ES-TIL samples have the lowest proportion of private (site-specific) clonotypes. This could possibly be due to expanded clonotypes ‘crowding out’ the read space for less-expanded, site-specific i.e. ‘private’ clonotypes in the ES-TIL samples (reviewer 3’s hypothesis would probably be this, based on their comments).