title: “Heart_deconvolution”
#https://github.com/rdong08/spatialDWLS_dataset
output: html_document

#https://github.com/rdong08/spatialDWLS_dataset/blob/main/codes/Heart_ST_tutorial_spatialDWLS.Rmd

##Import library

```{r}

library(Giotto)
library(ggplot2)
library(scatterpie)
library(data.table)
#getwd() path="G:/silicosis/sicosis/gitto/heart_visu_deconv/" setwd(path) ##Here is an example for tutorial by using spatial transcriptomic data of Sample 10, W6 in Asp et al. ##Identification of marker gene expression in single cell #{r}
#sc_meta<-read.table("…/datasets/heart_development_ST/sc_meta.txt",header = F,row.names = 1)
#sc_data<-read.table("…/datasets/heart_development_ST/all_cells_count_matrix_filtered.tsv",header = T,row.names = 1)

sc_meta=read.table(“G:/silicosis/sicosis/gitto/heart_visu_deconv/sc_meta.txt”,
header =FALSE,row.names = 1)
head(sc_meta)

sc_data<-read.table(“G:/silicosis/sicosis/gitto/heart_visu_deconv/all_cells_count_matrix_filtered.tsv/all_cells_count_matrix_filtered.tsv”,
header = T,row.names = 1)
head(sc_data)

my_python_path = “python path”
instrs = createGiottoInstructions(python_path = my_python_path)
heart_sc <- createGiottoObject(raw_exprs = sc_data,
instructions = instrs)

heart_sc <- normalizeGiotto(gobject = heart_sc, scalefactor = 6000, verbose = T)

heart_sc <- calculateHVG(gobject = heart_sc)

gene_metadata = fDataDT(heart_sc)
featgenes = gene_metadata[hvg == ‘yes’]$gene_ID

heart_sc <- runPCA(gobject = heart_sc, genes_to_use = featgenes, scale_unit = F)
signPCA(heart_sc, genes_to_use = featgenes, scale_unit = F)

#######calculate Sig for deconvolution, This step use DEG function implemented in Giotto

heart_sc@cell_metadata$leiden_clus <- as.character(sc_meta$V3)

scran_markers_subclusters = findMarkers_one_vs_all(gobject = heart_sc,
method = ‘scran’,
expression_values = ‘normalized’,
cluster_column = ‘leiden_clus’)

Sig_scran <- unique(scran_markers_subclustersgenes[which(scranmarkerssubclustersgenes[which(scran_markers_subclustersgenes[which(scranmarkerssubclustersranking <= 100)])

########Calculate median expression value of signature genes in each cell type

norm_exp<-2^(heart_sc@norm_expr)-1

id<-heart_sc@cell_metadata$leiden_clus
ExprSubset<-norm_exp[Sig_scran,]

Sig_exp<-NULL

for (i in unique(id)){
Sig_exp<-cbind(Sig_exp,(apply(ExprSubset,1,function(y) mean(y[which(id==i)]))))
}
colnames(Sig_exp)<-unique(id)
#```

##Spatial transcriptomic data analysis
#```{r}
##The heart spatial transcriptomic data is from Asp et al “A Spatiotemporal Organ-Wide Gene Expression and Cell Atlas of the Developing Human Heart”.
spatial_loc<-read.table(file="…/datasets/heart_development_ST/sample10_w6_loc.txt",header = F)

spatial_exp<-read.table(file="…/datasets/heart_development_ST/sample10_w6_exp.txt",header = T,row.names = 1)

##Transform ensemble gene to official gene name
ens2gene<-read.table(file="…/datasets/heart_development_ST/ens2symbol.txt",row.names = 1)

inter_ens<-intersect(rownames(spatial_exp),rownames(ens2gene))

filter_spatial_exp<-spatial_exp[inter_ens,]
rownames(filter_spatial_exp)<-as.character(ens2gene[inter_ens,])

##Generate Giotto objects and cluster spots
heart_w6_sample10_st <- createGiottoObject(raw_exprs = filter_spatial_exp,spatial_locs = spatial_loc,
instructions = instrs)
heart_w6_sample10_st <- filterGiotto(gobject = heart_w6_sample10_st,
expression_threshold = 1,
gene_det_in_min_cells = 10,
min_det_genes_per_cell = 200,
expression_values = c(‘raw’),
verbose = T)

heart_w6_sample10_st <- normalizeGiotto(gobject = heart_w6_sample10_st)
heart_w6_sample10_st <- calculateHVG(gobject = heart_w6_sample10_st)
gene_metadata = fDataDT(heart_w6_sample10_st)

featgenes = gene_metadata[hvg == ‘yes’]$gene_ID
heart_w6_sample10_st <- runPCA(gobject = heart_w6_sample10_st, genes_to_use = featgenes, scale_unit = F)
signPCA(heart_w6_sample10_st, genes_to_use = featgenes, scale_unit = F)

heart_w6_sample10_st <- createNearestNetwork(gobject = heart_w6_sample10_st, dimensions_to_use = 1:10, k = 10)
heart_w6_sample10_st <- doLeidenCluster(gobject = heart_w6_sample10_st, resolution = 0.4, n_iterations = 1000)
##Deconvolution based on signature gene expression and Giotto object
heart_w6_sample10_st <- runDWLSDeconv(gobject = heart_w6_sample10_st, sign_matrix = Sig_exp)
##The result for deconvolution is stored in heart_w6_sample10_st@spatial_enrichmentDWLS.Thefollowingcodesarevisualizationdeconvolutionresultsusingpieplotplotdata<−as.data.frame(heartw6sample10st@spatialenrichmentDWLS. The following codes are visualization deconvolution results using pie plot plot_data <- as.data.frame(heart_w6_sample10_st@spatial_enrichmentDWLS.Thefollowingcodesarevisualizationdeconvolutionresultsusingpieplotplotdata<−as.data.frame(heartw6sample10st@spatialenrichmentDWLS)[-1]
plot_col <- colnames(plot_data)

plot_datax<−as.numeric(as.character(heartw6sample10st@spatiallocsx <- as.numeric(as.character(heart_w6_sample10_st@spatial_locsx<−as.numeric(as.character(heartw6sample10st@spatiallocssdimx))
plot_datay<−as.numeric(as.character(heartw6sample10st@spatiallocsy <- as.numeric(as.character(heart_w6_sample10_st@spatial_locsy<−as.numeric(as.character(heartw6sample10st@spatiallocssdimy))
min_x <- min(plot_datax)plotdatax) plot_datax)plotdataradius <- 0.4

df <- data.frame()
p <- ggplot(df) + geom_point() + xlim(min(plot_datax)−1,max(plotdatax)-1, max(plot_datax)−1,max(plotdatax)+1) + ylim(min(plot_datay)−1,max(plotdatay)-1, max(plot_datay)−1,max(plotdatay)+1)
p + geom_scatterpie(aes(x=x, y=y, r=radius), data=plot_data, cols=plot_col, color=NA, alpha=.8) +
geom_scatterpie_legend(plot_data$radius, x=1, y=1) + theme_classic()
#``

Heart_deconvolution giotto解卷积相关推荐

Visium_Brain_deconvolution giotto解卷积
title: "Visium_Brain_deconvolution" output: html_document #https://github.com/rdong08/spat ...
深度全解卷积神经网络（附论文）
授权自AI科技大本营(ID: rgznai100) 本文共10315字,建议阅读10分钟. 本文帮你加深对卷积神经网络的理解,并对这个重要概念有一个全面的认知. [ 导读 ]深度卷积神经网络是这一波 ...
【 MATLAB 】z 变换中的卷积与解卷积
关于卷积的博文,之前也写过几篇: [ MATLAB ]conv 函数介绍(卷积和多项式乘法) [ MATLAB ]两个序列的卷积和运算的MATLAB实现(1) [ MATLAB ]两个序列的卷积和运算 ...
tensorflow学习笔记(三十二):conv2d_transpose (解卷积)
tensorflow学习笔记(三十二):conv2d_transpose ("解卷积") deconv解卷积,实际是叫做conv_transpose, conv_transpose ...
深度学习（二十七）可视化理解卷积神经网络（反池化 pooling unpooling）
不明白pooling,unpooling以及upsampling的区别,所以查阅了一些资料以下为转载原文可视化理解卷积神经网络原文地址:http://blog.csdn.net/hjimce/a ...
深度学习元老Yann Lecun详解卷积神经网络
深度学习元老Yann Lecun详解卷积神经网络本文联合编译:Blake.高斐雷锋网(公众号:雷锋网)注:卷积神经网络(Convolutional Neural Network)是一种前馈神 ...
深度学习（二十七）可视化理解卷积神经网络-ECCV 2014
可视化理解卷积神经网络原文地址:http://blog.csdn.net/hjimce/article/details/50544370 作者:hjimce 一.相关理论本篇博文主要讲解2014年 ...
DL之CNN：卷积神经网络算法简介之原理简介(步幅/填充/特征图)、七大层级结构(动态图详解卷积/池化+方块法理解卷积运算)、CNN各层作用及其可视化等之详细攻略
DL之CNN:卷积神经网络算法简介之原理简介(步幅/填充/特征图).七大层级结构(动态图详解卷积/池化+方块法理解卷积运算).CNN各层作用及其可视化等之详细攻略目录 CNN 的层级结构及相关概念 ...
python cnn模型_ZfNet解卷积：可视化CNN模型（ PythonCode可视化Cifar10）
前言由来已久,ANN方法被称为模式识别里面的"黑盒"方法,因为ANN模型不能使用明确的函数--形式化的数学公式进行表示,同时也意味着应对评价模型,面对函数寻求最优解的优化方程也不 ...

Heart_deconvolution giotto解卷积

title: “Heart_deconvolution”
#https://github.com/rdong08/spatialDWLS_dataset
output: html_document

```{r}

Heart_deconvolution giotto解卷积相关推荐

最新文章

热门文章

Heart_deconvolution giotto解卷积

title: “Heart_deconvolution” #https://github.com/rdong08/spatialDWLS_dataset output: html_document

```{r}

Heart_deconvolution giotto解卷积相关推荐

最新文章

热门文章

title: “Heart_deconvolution”
#https://github.com/rdong08/spatialDWLS_dataset
output: html_document