作者，Evil Genius
这一篇给大家推荐一个空转分析的优秀软件---SpaCET，文章在Estimation of cell lineages in tumors from spatial transcriptomics data，2023年1月发表于NC，个人感觉软件集中了空转分析的重要功能，包括细胞互作和共定位，软件在SpaCET。
SpaCET是一个R包，用于分析癌症空间转录组学(ST)数据集，以估计肿瘤微环境中的细胞谱系和细胞间相互作用。简单地说，SpaCET首先通过整合常见恶性肿瘤的基因模式来估计癌细胞的丰度。SpaCET随后使用约束回归模型校准局部组织密度，并确定基质细胞和免疫细胞谱系分数。此外，SpaCET可以揭示肿瘤微环境中假定的细胞-细胞相互作用。此外，SpaCET可以整合匹配的scRNA-seq数据集作为自定义参考来进行细胞类型反卷积。
安装
# install.packages("devtools")
devtools::install_github("data2intelligence/SpaCET")
示例一、依据内置数据集分析空间数据
创建对象（10X数据）
library(SpaCET)
# set the path to the in-house breast cancer ST data. User can set the paths to their own data.
visiumPath <- file.path(system.file(package = "SpaCET"), "extdata/Visium_BC")
# load ST data to create an SpaCET object.
SpaCET_obj <- create.SpaCET.object.10X(visiumPath = visiumPath)
# show this object.
str(SpaCET_obj)
# show this object.
SpaCET_obj@input$counts[1:8,1:6]
## 8 x 6 sparse Matrix of class "dgCMatrix"
## 50x102 59x19 14x94 47x13 73x43 61x97
## MIR1302-2HG . . . . . .
## FAM138A . . . . . .
## OR4F5 . . . . . .
## AL627309.1 . . . . . 1
## AL627309.3 . . . . . .
## AL627309.2 . . . . . .
## AL627309.4 . . . . . .
## AL732372.1 . . . . . .
展示基础信息
# calculate the QC metrics
SpaCET_obj <- SpaCET.quality.control(SpaCET_obj)
# plot the QC metrics
SpaCET.visualize.spatialFeature(
SpaCET_obj,
spatialType = "QualityControl",
spatialFeatures=c("UMI","Gene")
)
图片.png
Deconvolve ST data
通过两个阶段，SpaCET.deconvolution将所有ST spot的混合物分解为恶性细胞、免疫细胞和基质细胞。SpaCET首先通过拷贝数改变(CNA)的基因模式和常见恶性肿瘤的表达变化来估计恶性细胞的比例。随后，基于来自不同癌症类型的单细胞RNA-seq数据集的内部分层细胞谱系，使用约束回归模型来确定免疫和基质细胞分数，并包括一个无法识别的组件来校准跨组织区域的细胞密度变化。
具体来说，需要使用cancerType参数指定该肿瘤ST数据集中的癌症类型，利用该参数选择癌症类型特异性的CNA或表达特征来推断恶性细胞分数。癌症类型的代码可以在下拉列表中找到。当将ST点的表达谱与癌症类型特异性标记相关联时，将癌症类型特异性CNA标记作为第一选项，因为染色体不稳定性被广泛认为是人类肿瘤的一个一致特征。或者，如果没有点与CNA特征强相关，癌症类型特异性表达特征将被激活。这种情况可能是由于低CNA的染色体稳定的癌细胞。对于数据集中未包含的癌症类型，通过平均所有癌症类型特异性表达特征来创建泛癌症表达特征。
How many caner types are in the gene pattern dictionary?
# deconvolve ST data
SpaCET_obj <- SpaCET.deconvolution(SpaCET_obj, cancerType="BRCA", coreNo=8)
# show the ST deconvolution results
SpaCET_obj@results$deconvolution$propMat[1:13,1:6]
## 50x102 59x19 14x94 47x13 73x43
## Malignant 2.860636e-01 1 6.845966e-02 3.899756e-01 9.608802e-01
## CAF 3.118545e-01 0 3.397067e-01 1.111980e-01 3.372692e-02
## Endothelial 5.510895e-02 0 1.427060e-01 3.080531e-02 5.263544e-03
## Plasma 2.213392e-02 0 1.507382e-02 1.183170e-02 9.071809e-06
## B cell 3.885793e-03 0 9.271616e-02 1.406470e-01 1.329085e-06
## T CD4 1.344389e-01 0 1.554305e-02 1.249414e-01 1.112392e-05
## T CD8 7.578696e-03 0 2.514558e-07 1.379856e-03 1.123043e-06
## NK 7.104005e-04 0 1.670019e-06 4.890387e-08 3.562557e-07
## cDC 1.421632e-07 0 8.278023e-02 7.584295e-02 2.851146e-07
## pDC 1.606443e-06 0 2.283754e-02 1.805671e-02 3.878344e-07
## Macrophage 1.703304e-01 0 5.021248e-02 9.531511e-02 9.253645e-07
## Mast 7.905067e-08 0 1.621498e-05 1.333430e-07 1.162099e-07
## Neutrophil 1.380073e-05 0 9.528996e-07 1.167503e-08 9.908635e-05
可视化
# show the spatial distribution of malignant cells and macrophages.
SpaCET.visualize.spatialFeature(
SpaCET_obj,
spatialType = "CellFraction",
spatialFeatures=c("Malignant","Macrophage")
)
# show the spatial distribution of all cell types.
SpaCET.visualize.spatialFeature(
SpaCET_obj,
spatialType = "CellFraction",
spatialFeatures="All",
pointSize = 0.1,
nrow=5
)
图片.png
细胞互作
识别细胞共定位
SpaCET provides SpaCET.CCI.colocalization and SpaCET.visualize.colocalization to compute and visualize the co-localized cell-type pairs.
# calculate the cell-cell colocalization.
SpaCET_obj <- SpaCET.CCI.colocalization(SpaCET_obj)
# visualize the cell-cell colocalization.
SpaCET.visualize.colocalization(SpaCET_obj)
Analyze the L-R network enrichment within ST spots
细胞共定位并不直接表明物理相互作用。因此，我们通过分析ST点内配体-受体(L-R)的相互作用来寻求细胞-细胞相互作用的进一步证据。
# calculate the L-R network score across ST spots.
SpaCET_obj <- SpaCET.CCI.LRNetworkScore(SpaCET_obj,coreNo=8)
# visualize the L-R network score.
SpaCET.visualize.spatialFeature(
SpaCET_obj,
spatialType = "LRNetworkScore",
spatialFeatures=c("Network_Score","Network_Score_pv")
)
每个ST点的L-R网络评分表明了每个位置配体-受体相互作用的总体强度，而不是两种细胞类型之间的具体相互作用。因此，SpaCET进一步对每个细胞类型对进行了L-R网络评分的富集分析。例如，对于CAF和M2细胞在乳腺肿瘤组织中的共定位，SpaCET将所有ST点分为四类:CAF-M2共定位，CAF或M2占主导。分析发现，CAF- m2共定位点比CAF/ m2主导点具有更实质性的L-R交互网络评分。
Enrich cell-cell interactions at the tumor-immune interface
# Identify the Tumor-Stroma Interface
SpaCET_obj <- SpaCET.identify.interface(SpaCET_obj)
# Visualize the Interface
SpaCET.visualize.spatialFeature(SpaCET_obj, spatialType = "Interface")
# Compute the distance of CAF-M2 to tumor border
SpaCET.distance.to.interface(SpaCET_obj, cellTypePair=c("CAF", "Macrophage M2"))
Explore cancer cell states
肿瘤细胞的基因表达谱(即癌细胞状态)由肿瘤细胞的遗传背景和周围环境的细胞-细胞相互作用共同决定。作为探索性扩展，SpaCET可以自动探索不同癌细胞状态的空间分布。
# further deconvolve malignant cell states
SpaCET_obj <- SpaCET.deconvolution.malignant(SpaCET_obj, malignantCutoff = 0.7, coreNo = 8)
# show cancer cell state fraction of the first five spots
SpaCET_obj@results$deconvolution$propMat[c("Malignant cell state A","Malignant cell state B"),1:6]
## 50x102 59x19 14x94 47x13 73x43 61x97
## Malignant cell state A 0.2295498 9.999900e-01 6.845962e-02 0.2038680 9.608802e-01 0.6517794
## Malignant cell state B 0.0565137 1.239573e-11 3.921715e-08 0.1861075 2.340661e-09 0.2675332
SpaCET.visualize.spatialFeature(
SpaCET_obj,
spatialType = "CellFraction",
spatialFeatures=c("Malignant","Malignant cell state A","Malignant cell state B"),
nrow=1
)
示例二、依据匹配的单细胞数据集分析空间数据
library(SpaCET)
PDAC_Path <- system.file("extdata", 'oldST_PDAC', package = 'SpaCET')
load(paste0(PDAC_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)
解卷积
# load sc data
PDAC_Path <- system.file("extdata", 'oldST_PDAC', package = 'SpaCET')
load(paste0(PDAC_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"
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
)
基因空间表达可视化
# 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
)
网友评论