Celltalker

Connor H Knight

11/08/2021

Cell-cell communicaiton with celltalker

Installing celltalker

devtools::install_github("arc85/celltalker")

Running CellTalker

Celltalker is a simple tool for extracting cell-cell ligadn-receptor communication. In this tutorial we will show you how to run celltalker and produce a simple circos plot to visualise communication.

This command is simple loading in some example data and inserting it into an IBRAP object.

library(IBRAP)

system('curl -LJO https://raw.githubusercontent.com/connorhknight/IBRAP/main/data/celseq2.rds')

x <- RCurl::getURL('https://raw.githubusercontent.com/connorhknight/IBRAP/main/data/ScTypeDB.rds')

celseq2 <- readRDS('celseq2.rds')

obj <- createIBRAPobject(counts = celseq2$counts, original.project = 'celseq2', meta.data = celseq2$metadata) 

Here we are going to apply celltalker to our selected object. We can select counts and log-normalise the counts interally in the function. However, you also have the option to use already normalised counts.

results <- perform.celltalker(object = obj, assay = 'RAW', slot = 'counts', clust.method = 'metadata', column = 'celltype')

Visualisations

Now this is complete, we can filter the results by FDR-adjusted p-values to retain only the significant interactions. Next we can plot the interactions in a circos plot.

library(tidyverse)

top_stats <- results$results %>%
  mutate(fdr=p.adjust(p_val,method="fdr")) %>%
  filter(fdr<0.05) %>%
  group_by(cell_type1) %>%
  top_n(3,interact_ratio) %>%
  ungroup()

colors_use <- scales::hue_pal()(length(unique(obj[['celltype']])))

celltalker::circos_plot(ligand_receptor_frame=top_stats,
                        cell_group_colors=colors_use,
                        ligand_color="blue",
                        receptor_color="red",
                        cex_outer=0.5,
                        cex_inner=0.4)

Further information for celltalker can be found directly on their website: https://arc85.github.io/celltalker/index.html