upload

ZJUFanLab · Apr 5, 2022 · b441a3a · b441a3a
1 parent f962e66
commit b441a3a
Show file tree

Hide file tree

Showing 7 changed files with 999 additions and 204 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: SpaTalk
 Title: Infer Cell-Cell Communication for Spatial Transcriptomics
 Version: 1.0
-Depends: R (>= 4.0.0)
+Depends: R (>= 4.0.0), ggalluvial
 Authors@R: 
     person("Xin", "Shao", , "xin_shao@zju.edu.cn", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0002-1928-3878"))

diff --git a/R/methods.R b/R/methods.R
@@ -185,8 +185,7 @@ dec_celltype <- function(object, sc_data, sc_celltype, min_percent = 0.5, min_nF
     }
     st_ndata <- st_ndata[genename, ]
     sc_ndata <- sc_ndata[genename, ]
-    sc_celltype <- data.frame(cell = colnames(sc_ndata), celltype = sc_celltype,
-        stringsAsFactors = F)
+    sc_celltype <- data.frame(cell = colnames(sc_ndata), celltype = sc_celltype, stringsAsFactors = F)
     sc_celltype$celltype <- stringr::str_replace_all(sc_celltype$celltype, pattern = "-",
         replacement = "_")
     # if_use_all_cores
@@ -278,8 +277,8 @@ find_lr_path <- function(object, lrpairs, pathways, max_hop = NULL) {
     st_data <- .get_st_data(object)
     species <- object@para$species
     lrpair <- lrpairs[lrpairs$species == species, ]
-    lrpair <- lrpair[lrpair$ligand %in% rownames(st_data) & lrpair$receptor %in%
-        rownames(st_data), ]
+    lrpair <- lrpair[lrpair$ligand %in% rownames(st_data) & lrpair$receptor %in% rownames(st_data),
+        ]
     if (nrow(lrpair) == 0) {
         stop("No ligand-recepotor pairs found in st_data!")
     }
@@ -361,8 +360,8 @@ find_lr_path <- function(object, lrpairs, pathways, max_hop = NULL) {
 #' @importFrom crayon cyan green
 #' @export
 
-dec_cci <- function(object, celltype_sender, celltype_receiver, n_neighbor = 10,
-    min_pairs = 5, min_pairs_ratio = 0, per_num = 1000, pvalue = 0.05, co_exp_ratio = 0.1) {
+dec_cci <- function(object, celltype_sender, celltype_receiver, n_neighbor = 10, min_pairs = 5,
+    min_pairs_ratio = 0, per_num = 1000, pvalue = 0.05, co_exp_ratio = 0.1) {
     # check input data
     if (!is(object, "SpaTalk")) {
         stop("Invalid class for object: must be 'SpaTalk'!")
@@ -482,8 +481,8 @@ dec_cci_all <- function(object, n_neighbor = 10, min_pairs = 5, min_pairs_ratio
                 ### [2] Downstream targets and TFs
                 max_hop <- object@para$max_hop
                 if (nrow(lrdb) > 0) {
-                  receptor_tf <- .get_tf_res(celltype_sender, celltype_receiver,
-                    lrdb, ggi_tf, cell_pair, st_data, max_hop, co_exp_ratio)
+                  receptor_tf <- .get_tf_res(celltype_sender, celltype_receiver, lrdb,
+                    ggi_tf, cell_pair, st_data, max_hop, co_exp_ratio)
                   if (!is.null(receptor_tf)) {
                     # calculate score
                     lrdb <- .get_score(lrdb, receptor_tf)

diff --git a/R/plot.R b/R/plot.R
diff --git a/R/utils.R b/R/utils.R
@@ -375,8 +375,7 @@
         if (nrow(ggi_hop_yes) > 0) {
             ggi_hop_tf <- ggi_res[ggi_res$hop == k + 1, ]
             if (nrow(ggi_hop_tf) > 0) {
-                ggi_hop_yes <- ggi_hop_yes[ggi_hop_yes$dest %in% ggi_hop_tf$src,
-                  ]
+                ggi_hop_yes <- ggi_hop_yes[ggi_hop_yes$dest %in% ggi_hop_tf$src, ]
                 if (nrow(ggi_hop_yes) > 0) {
                   tf_gene <- ggi_hop_yes$hop
                   names(tf_gene) <- ggi_hop_yes$dest

diff --git a/vignettes/sc_tutorial.Rmd b/vignettes/sc_tutorial.Rmd
@@ -1,20 +1,20 @@
 ---
-title: "SpaTalk tutorial (single-cell ST data)"
+title: "SpaTalk tutorial (spot-based ST data)"
 author: "Xin Shao"
 date: "`r Sys.Date()`"
 output:
   prettydoc::html_pretty:
     theme: cayman
     highlight: github
 vignette: >
-  %\VignetteIndexEntry{SpaTalk tutorial (single-cell ST data)}
+  %\VignetteIndexEntry{SpaTalk tutorial (spot-based ST data)}
   %\VignetteEngine{knitr::rmarkdown}
   %\VignetteEncoding{UTF-8}
 ---
 
-## **Preprocess**
+## **Main**
 
-Before create SpaTalk object, we suggest to revise gene symbols of ST data `matrix` and scRNA-seq data `matrix` using `rev_gene()`
+After pre-processing, we create a SpaTalk object with st_data and st_meta for single-cell ST data by setting `if_st_is_sc` and `spot_max_cell` as `TRUE` and `1`
 
 ```{r load st data, echo=TRUE}
 library(SpaTalk)
@@ -23,45 +23,196 @@ library(SpaTalk)
 load(paste0(system.file(package = 'SpaTalk'), "/extdata/starmap_data.rda"))
 load(paste0(system.file(package = 'SpaTalk'), "/extdata/starmap_meta.rda"))
 
-# revise gene symbols according to the NCNI gene symbols
-starmap_data <- rev_gene(data = as.matrix(starmap_data),
-                         data_type = "count",
-                         species = "Mouse",
-                         geneinfo = geneinfo)
+# create SpaTalk data
+obj <- createSpaTalk(st_data = as.matrix(starmap_data),
+                     st_meta = starmap_meta[, -4],
+                     species = "Mouse",
+                     if_st_is_sc = T,
+                     spot_max_cell = 1)
 
 ```
 
+## **Cell-type decomposition**
 
-## **Note**
+First, we apply [NNLM](https://github.com/linxihui/NNLM) and spatial mapping to reconstruct the ST data at single-cell resolution using `dec_celltype()`
+
+```{r cell-type decomposition, echo=TRUE}
+# decode the cell-type composition
+obj <- dec_celltype(object = obj,
+                    sc_data = as.matrix(starmap_data),
+                    sc_celltype = starmap_meta$celltype)
+
+```
+
+### Use `plot_st_celltype_percent()` to view cell-type percent
+
+```{r plot_st_celltype_percent, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot cell-type percent across spatial cells
+plot_st_celltype_percent(object = obj, celltype = 'eL2_3',size = 3)
+
+```
+
+### Use `plot_st_gene()` to view gene expression
+
+```{r plot_st_gene, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot marker gene expression across spatial cells
+plot_st_gene(object = obj, gene = 'Lamp5',size = 3, if_use_newmeta = F)
+
+```
+
+### Use `plot_st_cor_heatmap()` to view correlation heatmap
+
+```{r plot_st_cor_heatmap, fig.width=7, fig.height=6, fig.align='center', echo=TRUE}
+# correlation between marker gene expression and cell type percent across spatial cells
+plot_st_cor_heatmap(object = obj,
+                    marker_genes = c("Plp1","Vip","Sst","Lamp5","Pcp4","Mfge8","Pvalb"),
+                    celltypes = c("Oligo","VIP","SST","eL2_3","eL6","Astro","PVALB"),
+                    scale = "none",
+                    color_low = 'blue',
+                    color_high = 'yellow',
+                    color_mid = 'yellow')
+```
 
-In order to guide users on how to create the right data type, we provide several `demo_data` to follow:
+## **ST at single-cell resolution**
+### Use `plot_st_celltype()` to view cell type in space
 
-```{r demo data, echo=TRUE}
-# demo spot-based ST data
-demo_st_data()
+```{r plot_st_celltype, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot cell type in reconstructed ST atlas
+plot_st_celltype(object = obj, celltype = 'eL2_3')
 
-# demo spot-based ST meta
-demo_st_meta()
+```
 
-# demo single-cell ST data
-demo_st_sc_data()
+### Use `plot_st_gene()` to view gene expression in space
 
-# demo single-cell ST meta
-demo_st_sc_meta()
+```{r plot_st_gene2, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot marker gene expression in single-cell ST data
+plot_st_gene(object = obj,gene = 'Lamp5', if_use_newmeta = T)
 
-# demo scRNA-seq data
-demo_sc_data()
+```
 
-# demo geneinfo
-demo_geneinfo()
+### Use `plot_st_celltype_density()` to view cell-type density in space
+
+```{r plot_st_celltype_density, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot cell-type density in single-cell ST data
+plot_st_celltype_density(object = obj,
+                         celltype = 'eL2_3',
+                         type = 'raster',
+                         color_low = 'purple',
+                         color_high = 'yellow')
+
+plot_st_celltype_density(object = obj,
+                         celltype = 'eL2_3',
+                         type = 'contour',
+                         color_low = 'purple',
+                         color_high = 'yellow')
+```
+
+### Use `plot_st_celltype_all()` to view all cell types in space
+
+```{r plot_st_celltype_all, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# plot all cell types in single-cell ST data
+plot_st_celltype_all(object = obj,size = 2)
 
 ```
 
-After revising gene symbols, we create a SpaTalk object with st_data and st_meta. Given the two types of ST data, namely single-cell and spot-based ST data, we provided two standard processing examples. Please refer to:
+## **Infer cell-cell communications**
+Based on the single-cell ST data, we use `find_lr_path()` and `dec_cci()` to infer cell-cell communications mediated by ligand-receptor interactions (LRIs) in space using [LRIs from CellTalkDB](http://tcm.zju.edu.cn/celltalkdb/), pathways from [KEGG](https://www.genome.jp/kegg/pathway.html) and [Reactome](https://reactome.org/), and transcriptional factors (TFs) from [AnimalTFDB](http://bioinfo.life.hust.edu.cn/AnimalTFDB/#!/). In practise, we use the real cell type of STARmap
+
+```{r dec_cci, echo=TRUE}
+obj@meta$rawmeta$celltype <- starmap_meta$celltype
+# Filter LRIs with downstream targets
+obj <- find_lr_path(object = obj, lrpairs = lrpairs, pathways = pathways)
 
-## [**`single-cell tutorial vignette`**](https://raw.githack.com/ZJUFanLab/SpaTalk/main/vignettes/sc_tutorial.pdf)
+# Infer cell-cell communications from SST to PVALB neurons
+obj <- dec_cci(object = obj, 
+               celltype_sender = 'eL5',
+               celltype_receiver = 'Astro')
 
-## [**`spot-based tutorial vignette`**](https://raw.githack.com/ZJUFanLab/SpaTalk/main/vignettes/spot_tutorial.pdf)
+# Get LR and downstream pathways
+obj_lr_path <- get_lr_path(object = obj,
+                           celltype_sender = 'eL5',
+                           celltype_receiver = 'Astro',
+                           ligand = 'Cort',
+                           receptor = 'Sstr2')
+
+obj_lr_path$tf_path
+obj_lr_path$path_pvalue
+```
+
+### Use `plot_ccdist()` to view distribution of senders and receivers
+
+```{r plot_ccdist, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# Point plot with spatial distribution of celltype_sender and celltype_receiver
+plot_ccdist(object = obj, celltype_sender = 'eL5', celltype_receiver = 'Astro')
+
+```
+
+### Use `plot_cci_lrpairs()` to view all LRIs of senders and receivers
+
+```{r plot_cci_lrpairs, fig.width=3, fig.height=3, fig.align='center', echo=TRUE}
+# Heatmap with LR pairs of celltype_sender and celltype_receiver
+plot_cci_lrpairs(object = obj, celltype_sender = 'eL5', celltype_receiver = 'Astro')
+
+```
+
+### Use `plot_lrpair()` to view the specific LRI
+
+```{r plot_lrpair, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# Point plot with LR pair from celltype_sender to celltype_receiver
+plot_lrpair(object = obj,
+            celltype_sender = 'eL5',
+            ligand = 'Cort',
+            celltype_receiver = 'Astro',
+            receptor = 'Sstr2',
+            if_plot_density = F)
+
+```
+
+### Use `plot_lrpair_vln()` to view violin plot with spatial distance of LRI
+
+```{r plot_lrpair_vln, fig.width=8, fig.height=6, fig.align='center', echo=TRUE}
+# Violin plot with spatial distance of LR pair between senders and receivers and between all cell-cell pairs
+plot_lrpair_vln(object = obj,
+                celltype_sender = 'eL5',
+                ligand = 'Cort',
+                celltype_receiver = 'Astro',
+                receptor = 'Sstr2')
+
+```
+
+### Use `plot_lr_path()` to view network with LR and downstream pathways
+
+```{r plot_lr_path, fig.width=6, fig.height=6, fig.align='center', echo=TRUE}
+# Plot network with LR and downstream pathways
+plot_lr_path(object = obj,                
+             celltype_sender = 'eL5',
+             ligand = 'Cort',
+             celltype_receiver = 'Astro',
+             receptor = 'Sstr2')
+
+```
+
+### Use `plot_path2gene()` to view river plot of significantly activated pathways
+
+```{r plot_path2gene, fig.width=8, fig.height=6, fig.align='center', echo=TRUE}
+# River plot of significantly activated pathways and related downstream genes of receptors
+library(ggalluvial)
+plot_path2gene(object = obj,                
+               celltype_sender = 'eL5',
+               ligand = 'Cort',
+               celltype_receiver = 'Astro',
+               receptor = 'Sstr2')
+
+```
+
+## **Note**
+
+To infer all paired cell-cell communications, use `dec_cci_all()` instead of `dec_cci()`
+
+```{r dec_cci_all, echo=TRUE}
+# Infer cell-cell communications from SST to PVALB neurons
+# obj <- dec_cci_all(object = obj)
+```
 
 ```{r sessionInfo, echo=TRUE}
 sessionInfo()