TimeTalk An R package to perform cell-cell communication studies by utilizing temporal information behind single-cell sequencing, which is suitable for development biology research. In brief, TimeTalk uses the temporal information unraveled by trajectory analysis and the Granger causal test to test the ligand-receptor pairs with causal relationships to master transcription factors during development. Therefore it can elucidate the potential causal relationship between ligand-receptor interaction and TF activity.
Require systems: Ubuntu (20.04), users can test other linux distributions.
- Install require software in Ubuntu:
sudo apt-get install libcairo2-dev
- Install require package in R
devtools::install_github('cole-trapnell-lab/monocle3')
devtools::install_github("shenorrLab/cellAlign")
BiocManager::install("RTN")
devtools::install_github("ChengLiLab/TimeTalk",ref="main")
Users can run this demo to quikstart of TimeTalk
library(TimeTalk)
library(Seurat)
library(tidyverse)
library(monocle3)
tmp.path <- system.file("extdata/seu_demo.rds",package = "TimeTalk")
seu.small <- readRDS(file = tmp.path)
tmp.path <- system.file("extdata/cds_demo.rds",package = "TimeTalk")
cds.small <- readRDS(file = tmp.path)
tmp.path <- system.file("extdata/B_blastoid_RTN_mra_result.rds",
package = "TimeTalk")
tmp.mra.res <- readRDS(file = tmp.path)
tmp.path <- system.file("extdata/Mouse-2020-Shao-LR-pairs.txt",
package = "TimeTalk")
LRpairs.df <- read.delim(file = tmp.path,
stringsAsFactors = F)
TimeTalk.result <- RunTimeTalk(tmp.cds=cds.small,
tmp.seu=seu.small,
tmp.orig.ident = "blastocyst",
tmp.ident.1="EPI",
tmp.ident.2 = "PE",
LRpairs.df = LRpairs.df,
tmp.mra.res = tmp.mra.res,
tmp.winsz = 0.1,
tmp.lags = 1,
numPts = 200,
tmp.SCC.cutoff = 0.2,
tmp.granger.cutoff = 1e-2)
tmp.res <- TimeTalk.result %>%
filter(category == "PASS") %>%
pull(LR) %>%
unique()
As for demonstration, we don't recommend users to run this code, as it was very time consuming. But we encourage the users to test this code for their own data.
library(TimeTalk)
### build all the CellType specific TRN,
### Before run it make sure it will take too long time
lapply(levels(seu), function(ii){
rtna <- myGetCellTypeSpecificTRN(tmp.seu = seu,
tmp.ident = ii,
gene.TF = gene.TF,
tmp.phenotype = tmp.phenotype)
saveRDS(rtna,file = myFileName(prefix = paste0("res/R/",ii,"RTN_rtna_object"),suffix = ".rds"))
})
### not run again!!!
tmp.all.de.RNA <- FindAllMarkers(object = seu,
assay = "RNA",
only.pos = T)
####read result
tmp.res.list <- lapply(levels(seu),FUN = function(ii){
cat(ii,sep = "\n")
tmp.ident <- ii
tmp.files <- list.files(path = "res/R",pattern = paste0(tmp.ident,"RTN_rtna_object"))
rtna <- readRDS(file = paste0("res/R/",tmp.files))
mra <- tna.get(rtna, what="mra", ntop = -1)
tmp.gene.use <- tmp.all.de.RNA %>%
filter(cluster == tmp.ident) %>%
filter(p_val_adj < 0.05) %>%
pull(gene)
mra.res <- mra %>%
filter(Pvalue < 0.05) %>%
filter(Regulon %in% tmp.gene.use) %>%
mutate(group = tmp.ident)
return(mra.res)
})
tmp.mra.res <- Reduce(rbind,tmp.res.list)
saveRDS(object = tmp.mra.res,file = "res/R/B_blastoid_RTN_mra_result.rds")
tmp.mra.res <- readRDS(file = "res/R/B_blastoid_RTN_mra_result.rds")
LRpairs.df <- read.delim(file = "database/Ligand-Receptor-Pairs/Mouse/Mouse-2020-Shao-LR-pairs.txt",stringsAsFactors = F)
TimeTalk.result <- RunTimeTalk(tmp.cds=tmp.cds,
tmp.seu=tmp.seu,
tmp.ident.1="EPI",
tmp.ident.2 = "PE",
LRpairs.df = LRpairs.df,
tmp.mra.res = tmp.mra.res,tmp.winsz = 0.1,
numPts = 200,
tmp.SCC.cutoff = 0.2,
tmp.granger.cutoff = 1e-2)
According to our simulation results, the Granger causality conclusion is affected by choice of parameter winsz. Thus, we acknowledge the need for additional methods to validate the inferred potential regulatory relationship between ligand-receptor interactions and transcription factors. Regrettably, the scarcity of high temporal resolution datasets hinders our ability to address this issue. Despite this limitation, we have acknowledged in the Discussion section of the manuscript that our causal analysis serves as a helpful reference, and we intend to improve our approach in the future with more accurate data. Furthermore, we advise users to conduct additional data analysis and experiment with various parameter combinations to achieve more dependable results when utilizing TimeTalk for cell communication analysis.
-
update log and fix function bugs: 2023.05.02.
-
update the documentation file: 2023.06.01.
-
Add the demo data: 2023.07.30
Wang, L., Zheng, Y., Sun, Y. et al. TimeTalk uses single-cell RNA-seq datasets to decipher cell-cell communication during early embryo development. Commun Biol 6, 901 (2023). https://doi.org/10.1038/s42003-023-05283-2