Skip to contents

Deconvolution with a matched scRNA-seq data set

Source: vignettes/oldST_PDAC.Rmd
oldST_PDAC.Rmd

Since most tumor spatial transcriptomics (ST) data do not have matched scRNA-seq data from the same sample, SpaCET does not require malignant, stromal and immune cell reference. However, SpaCET can still accept a customized reference to carry out cell type deconvolution. This tutorial demonstrates how to run SpaCET with a matched scRNA-seq dataset by using a pancreatic ductal adenocarcinoma (PDAC) ST data set from Moncada et al, 2020. Each spot is 100 µm in diameter.

Create SpaCET object

To read your ST data into R, user can create an SpaCET object by using create.SpaCET.object or create.SpaCET.object.10X. Specifically, if users are analyzing an ST dataset from 10x Visium, they only need to input “visiumPath” by using create.SpaCET.object.10X. Please make sure that “visiumPath” points to the standard output folders of 10x SpaCET Ranger, which has both “filtered_feature_bc_matrix” and “spatial” folders.

Here, since the PDAC ST data set was generated from the original ST technology, user need create.SpaCET.object to create an SpaCET object by preparing four types of input data referring to a tumor ST sample.

  1. spatial transcriptomics count data. The spatial transcriptomics count data must be in the format of matrix with gene name (row) x spot ID (column).
  2. spatial location information. The spot coordinates should be in the format of matrix with spot ID (row) x coordinates (column). This 1st and 2nd columns represent X and Y coordinates, respectively.
  3. path to the H&E image file. The image path can be NA if unavailable.
  4. platform.
library(SpaCET)

oldST_Path <- system.file("extdata", 'oldST_PDAC', package = 'SpaCET')
load(paste0(oldST_Path,"/st_PDAC.rda"))

# show count matrix
counts[1:6,1:5]

##        10x10 10x13 10x14 10x15 10x16
## A1CF       0     0     0     0     0
## A2M       13     0     4     0     0
## A4GALT     1     0     0     0     0
## A4GNT      0     0     1     0     0
## AAAS       0     0     0     0     0
## AACS       0     0     0     0     0

# show coordinate matrix
spotCoordinates[1:5,]

##        X  Y
## 10x10 10 10
## 10x13 10 13
## 10x14 10 14
## 10x15 10 15
## 10x16 10 16

# load ST data to create an SpaCET object.
SpaCET_obj <- create.SpaCET.object(
  counts=counts,
  spotCoordinates=spotCoordinates,
  imagePath=NA,
  platform = "oldST"
)

# show this object.
str(SpaCET_obj)

Deconvolve cell lineage

We provide SpaCET.deconvolution.matched.scRNAseq to deconvolve an SpaCET object with a customized scRNA-seq data. User need to prepare three types of input data referring to the matched scRNA-seq dataset.

  1. single cell RNA-seq (scRNA-seq) count data. The scRNA-seq count data must be in the format of matrix with gene name (row) x cell ID (column).
  2. cell annotation information. This matrix should include two columns, i,e., cellID and cellType. Each row represents a single cell. The name of row should be same as the column cellID.
  3. Hierarchical tree of cell types. This should be organized by using a list, and the name of each element are major lineages while the value of elements are the corresponding sublineages. If a major lineage does not have any sublineages, the value of this major lineage should be itself.
# load sc data
oldST_Path <- system.file("extdata", 'oldST_PDAC', package = 'SpaCET')
load(paste0(oldST_Path,"/sc_PDAC.rda"))

# show count matrix
sc_counts[1:6,1:5]

##         c1 c2 c3 c4 c5
## A1BG     0  0  0  0  0
## A1CF     0  0  0  1  0
## A2M      0  0  0  0  0
## A2ML1    0  0  0  0  0
## A3GALT2  0  0  0  0  0
## A4GALT   0  0  0  0  0

# show cell annotation matrix
sc_annotation[1:6,]

##    cellID bio_celltype                            
## c1 "c1"   "Acinar cells"                          
## c2 "c2"   "Ductal - terminal ductal like"         
## c3 "c3"   "Ductal - terminal ductal like"         
## c4 "c4"   "Ductal - CRISP3 high/centroacinar like"
## c5 "c5"   "Cancer clone A"                        
## c6 "c6"   "Cancer clone A" 

# show cell type lineage tree
head(sc_lineageTree)

## $Cancer
## [1] "Cancer clone A" "Cancer clone B"
## 
## $Ductal
## [1] "Ductal - APOL1 high/hypoxic"            "Ductal - CRISP3 high/centroacinar like"
## [3] "Ductal - MHC Class II"                  "Ductal - terminal ductal like"         
## 
## $Macrophage
## [1] "Macrophages A" "Macrophages B"
## 
## $mDC
## [1] "mDCs A" "mDCs B"
## 
## $`Acinar cells`
## [1] "Acinar cells"
## 
## $`Endocrine cells`
## [1] "Endocrine cells"

Then, user can run SpaCET.deconvolution.matched.scRNAseq to carry out cell type deconvolution.

SpaCET_obj <- SpaCET.deconvolution.matched.scRNAseq(
  SpaCET_obj, 
  sc_counts=sc_counts, 
  sc_annotation=sc_annotation, 
  sc_lineageTree=sc_lineageTree, 
  coreNo=8
)

SpaCET.visualize.spatialFeature(
  SpaCET_obj, 
  spatialType = "CellFraction",
  spatialFeatures = c("Cancer clone A","Cancer clone B","Acinar cells","Ductal - CRISP3 high/centroacinar like"),
  nrow=2
)

User can use the following code to visualize the marker gene expression level and verify the cell type deconvolution.

# Markers for cancer clone A and B, acinar cell, and centroacinar like ductal cell
SpaCET.visualize.spatialFeature(
  SpaCET_obj,
  spatialType = "GeneExpression",
  spatialFeatures = c("TM4SF1","S100A4","PRSS1","CRISP3"),
  nrow=2
)

Extand this function module

Although our tutorials focus on analyzing ST data from tumor samples, in principle, SpaCET can be applied to any ST datasets with matched scRNA-seq data.