ixxmu
commented
11 months ago 复现Nature经典空转Spatial multiomic map of human myocardial infarction by 生信菜鸟团
个人感觉这篇文章在空转领域太经典了。在更高精度的空转技术出来之前，这篇技术非常值得学习。因文章本身代码太多，最近刚理清了他的代码思路，同时结合自己的数据，就开始着手复现。如果感兴趣可以跟我一起学习，估计这篇文章完整复现需要1年时间
，敬请期待。
文章的复现，需要使用cell2location，所以要了解hpc高性能计算，学会使用bsub提交作业。或者你可以使用其他软件获取细胞丰度文件也行。最重要的是理解Misty这个r包的原理和使用。
复现分为两部分：
是关于本部分必须掌握的基础知识，如果你连基础知识都没掌握，就没必要看第二步了哦
是最终的代码部分，有的地方我会加上注释
提醒：这部分的复现要求：1需要你已经制备好seurat对象的空转文件，里面包含矩阵和image才能正常运行哦。2你要学会progeny包的使用 3 如果有细胞丰度文件就更好。
01
purr是一个R语言中的包，旨在简化数据操作的过程，提供了一组功能强大且一致的工具，用于对列表数据进行操作，但它的并行处理是指每个输入同时处理，不涉及到多核计算。
这里是官方提到的参数及其含义：
https://purrr.tidyverse.org/reference/pmap.html下面这个链接可以加深我们对r中向量 数据框 列表的理解：
https://dcl-prog.stanford.edu/data-structures.html02
options(warn = -1)#request 2.libPaths(c( "/home/data/t040413/R/x86_64-pc-linux-gnu-library/4.2","/home/data/t040413/R/yll/usr/local/lib/R/site-library","/home/data/refdir/Rlib/", "/usr/local/lib/R/library"))#https://github.com/saezlab/misty_pipelines/blob/0d23e52524375171bd26182892a18fc55b258818/visium/2_run_misty_ppline.R#L30#https://github.com/saezlab/visium_heart#https://saezlab.github.io/mistyR/articles/# install.packages("remotes")#remotes::install_github("saezlab/mistyR")library(conflicted)conflicts_prefer(dplyr::filter())conflicts_prefer(dplyr::lag())library(tidyverse)library(Seurat)library(progeny)library(conflicted)conflicts_prefer(dplyr::filter())conflicts_prefer(dplyr::lag())library(conflicted)conflict_scout()print(getwd())dir.create("~/silicosis/spatial/prcessed_visum_for_progeny/",recursive = TRUE)setwd("~/silicosis/spatial/prcessed_visum_for_progeny/")#' @param visium_slide: Seurat object with SCT assay#' @param species: human or mouse, will extract regulons based on this#' @param top: number of genes used in footprint#' @return A Seurat object with pathway activities in progeny assayadd_path_activities <- function(visium_slide,species = "human",top = 500,assay = "RNA"){if(species == "mouse"){model <- progeny::getModel(organism = "Mouse", top = top)common_genes <- intersect(rownames(GetAssayData(visium_slide, assay = assay)), rownames(model))progeny_scores <- (t(model)[, common_genes]) %*% (GetAssayData(visium_slide, assay = assay)[common_genes, ])# Here we scale by featureprogeny_scores <- scale(t(as.matrix(progeny_scores)))visium_slide[['progeny']] <- CreateAssayObject(counts = t(progeny_scores))visium_slide@assays# visium_slide <- readRDS("./processed_visium/SiO2_56.rds")# marker.list=read.table("~/silicosis/markers_for_each_silicosis.csv",header = TRUE,sep = ",")# marker.list=read.table("/home/data/t040413/silicosis/data/tabula_scRNAseq/integration_with_sc_silicosis/markers__foreach_10x.csv",header = TRUE,sep = ",")# marker.list=read.csv("/home/data/t040413/silicosis/markers_foreach_silico_fibro.csv")# load("~/silicosis/silicos_fibro_special_grem1_inflamm.rds" )# load("~/silicosis/sc_marker.list_macro.rds")# #markers_for_Each_macro=read.csv("~/silicosis/markers_for_Each_macro_silicosi.csv")## marker.list=read.csv("/home/data/t040413/silicosis/marker.list_silicosis_fibro_AM3_mappedbacked.csv")### head(marker.list)# sc_marker.list=list()## for (each in unique(marker.list$cluster)) {# sc_marker.list[[each]]=marker.list[marker.list$cluster==each & marker.list$avg_log2FC>1 & marker.list$p_val_adj<0.05,"gene"]# }### sc_marker.list=c(# sc_marker.list_fibro,# sc_marker.list[!c( names(sc_marker.list) %in% names(sc_marker.list_macro) )] ,# sc_marker.list_macro# )### head(visium_slide@meta.data)# # assay="Spatial"# visium_slide=AddModuleScore(visium_slide,features =sc_marker.list ,assay = assay)## colnames(visium_slide@meta.data)# visium_slide@meta.data[ ,(ncol(visium_slide@meta.data)-length(sc_marker.list)+1 ): ncol(visium_slide@meta.data)] %>%head()# sc_marker_names=names(sc_marker.list)## colnames(visium_slide@meta.data)[ (ncol(visium_slide@meta.data)-length(sc_marker.list)+1 ): ncol(visium_slide@meta.data)] =sc_marker_names# module_score=FetchData(visium_slide,vars = sc_marker_names,)# print(dim(module_score))# dim(visium_slide)# # head(module_score)[,1:9]# # Here we scale by feature# module_score <- scale(as.matrix(module_score))# visium_slide@[['module_score']] <- CreateAssayObject(counts = t(module_score))# visium_slide[['Spatial']]# visium_slide <- readRDS("./processed_visium/SiO2_56.rds")visium_slide@meta.data %>%head()visium_slide$aling_with_spatial= paste(visium_slide$orig.ident,sep = "_",gsub(pattern = "_[1-4]",replacement = "",x = colnames(visium_slide)))visium_slide@meta.data %>%head()cell_abundance=read.csv("/home/data/t040413/silicosis/spatial_c2l_results/silicosismappedback_noinflammatory/q05_cell_abundance_w_sf.csv",header = TRUE,row.names = "spot_id")head(cell_abundance)table(rownames(cell_abundance) %in% visium_slide$aling_with_spatial)# # Here we scale by spotcell_abundance_scales= apply(cell_abundance, 2, function(x) (x - min(x)) / (max(x) - min(x)))########为了让不同点之间具有可比性，每列scalehead(cell_abundance_scales)cell_abundance_scales["SiO2_56_AAACAAGTATCTCCCA-1",]tmp= cell_abundance[rownames( cell_abundance) %in% visium_slide$aling_with_spatial,]head(tmp)[,1:9]head(colnames(visium_slide))head(visium_slide@meta.data)rownames(tmp)=colnames(visium_slide)visium_slide= AddMetaData(visium_slide,metadata = tmp)head(visium_slide@meta.data)tmp2=FetchData(visium_slide,vars = colnames(tmp))visium_slide@assays[["cell_abundance_scaled"]]= CreateSeuratObject(counts = t( visium_slide@meta.data[,colnames(tmp)] )) %>%GetAssay(assay="RNA")visium_slide@assaysvisium_slide[["cell_abundance_scaled"]]}else if(species == "human"){model <- progeny::getModel(organism = "Human", top = top)common_genes <- intersect(rownames(GetAssayData(visium_slide, assay = assay)), rownames(model))progeny_scores <- (t(model)[, common_genes]) %*% (GetAssayData(visium_slide, assay = assay)[common_genes, ])# Here we scale by featureprogeny_scores <- scale(t(as.matrix(progeny_scores)))visium_slide[['progeny']] <- CreateAssayObject(counts = t(progeny_scores))}return(visium_slide)}#setwd("../")print(getwd())## Process all datadir.create("processed_visium")dir.create("processed_visium_3")param_df <- tibble("sample_name" = list.dirs("./data",full.names = F, recursive = F),"data_dir" = list.dirs("./data",full.names = T, recursive = F),"sample_file_self"=normalizePath(list.files("./data/", recursive = TRUE, full.names = TRUE) )) %>%dplyr::mutate("out_folder" = paste0("./processed_visium_3/", sample_name, ".rds"))param_dfprint(getwd())#normalizePath(list.files(path = file.path("./data",sample_name), recursive = TRUE, full.names = TRUE) )# fs::dir_tree()# fs::dir_ls("data")purrr::walk2(param_df$data_dir, param_df$out_folder, function(dir_name,out_dir_name) {# dir_name="./data/NS_56"# samplefile_self="/home/data/t040413/silicosis/spatial/prcessed_visum_for_progeny/data/NS_56/NS_56.rds"# out_dir_name="./processed_visium/NS_56.rds"print(dir_name)sample_name <- gsub("./data/", "", dir_name)sample_file_self= normalizePath( list.files( path = file.path("./data",sample_name), recursive = TRUE, full.names = TRUE) )visium_slide = readRDS(paste0(sample_file_self)) %>%SCTransform(. , assay = "Spatial", verbose = F) %>%NormalizeData(.,normalization.method = 'LogNormalize',scale.factor = 10000,verbose = FALSE) %>%add_path_activities(.,species = "mouse",#"human",top = 1000,assay = "SCT")tmplist= visium_slide@images[[sample_name]]visium_slide@images=list()visium_slide@images= list(slice1=tmplist)# str(visium_slide)# str(visium_slide@images[[sample_name]])# saveRDS(visium_slide, file = out_dir_name)# ##########1-------# DefaultAssay(visium_slide) <- 'module_score'# # sample_name="sio2_56"# dir.create("~/silicosis/spatial/prcessed_visum_for_progeny/results/module_spatial_plts/",recursive = TRUE)# pdf(paste0("~/silicosis/spatial/prcessed_visum_for_progeny/results/module_spatial_plts/", sample_name, ".pdf"), height = 12, width = 10)## all_paths <- SpatialFeaturePlot(visium_slide,# features =rownames(visium_slide),# ncol = 4,# stroke = 0,# max.cutoff = "q99")+ggtitle("")# plot(all_paths)## dev.off()# pdf(paste0("~/silicosis/spatial/prcessed_visum_for_progeny/results/module_spatial_plts/", sample_name, "_2.pdf"), height = 12, width = 10)# for (each in rownames(visium_slide)) {# p=SpatialFeaturePlot(visium_slide,# features =each,# ncol = 4,# stroke = 0,# max.cutoff = "q99")# plot(p)### }# dev.off()############2-----DefaultAssay(visium_slide) <- "progeny"dir.create("./results/progeny_spatial_plts/",recursive = TRUE)pdf(paste0("./results/progeny_spatial_plts/", sample_name, ".pdf"), height = 12, width = 10)all_paths <- SpatialFeaturePlot(visium_slide,features = rownames(visium_slide),ncol = 4,stroke = 0,max.cutoff = "q99")plot(all_paths)dev.off()print(paste0(getwd(),sample_name,"=====sample_=========done------"))##########3----DefaultAssay(visium_slide) <- "cell_abundance_scaled"dir.create("./results/progeny_spatial_plts/cell_abundance_scaled",recursive = TRUE)pdf(paste0("./results/progeny_spatial_plts/cell_abundance_scaled", sample_name, "_.pdf"), height = 12, width = 10)all_paths <- SpatialFeaturePlot(visium_slide,features = rownames(visium_slide),ncol = 2,stroke = 0,max.cutoff = "q99")plot(all_paths)dev.off()######3.2----pdf(paste0("~/silicosis/spatial/prcessed_visum_for_progeny/results/module_spatial_plts/", sample_name, "_3.pdf"), height =12, width = 10)for (each in rownames(visium_slide)) {p=SpatialFeaturePlot(visium_slide,features =each,ncol = 2,stroke = 0,max.cutoff = "q99")plot(p)p=SpatialFeaturePlot(visium_slide,features =each,max.cutoff = "q99")plot(p)}dev.off()})
https://mp.weixin.qq.com/s/EgHYBTXUUSjW3SUv_dd1kQ