两年了还是没有R包，所以就得在Linux操作啦，而这在Windows的RStudio操作是满屏的error呀！

先Git clone一下，总是下载不了：git clone https://github.com/edroaldo/cellrouter.git

看来只能弄一下wget直接下载压缩包了，多次尝试都失败，下载的文件不完整：wget https://github.com/edroaldo/cellrouter/archive/master.zip

最终还是git clone搞定，大小也匹配：

https://static-content.springer.com/esm/art%3A10.1038%2Fs41467-018-03214-y/MediaObjects/41467_2018_3214_MOESM14_ESM.txt

先一步一步运行作者的tutorial：

edroaldo/cellrouter: Reconstruction of complex single-cell trajectories using CellRouter

https://github.com/edroaldo/cellrouter

list.of.packages 'reshape',new.packages list.of.packages[!(if(length(new.packages)) install.packages(new.packages, repos=c("http://cran.rstudio.com/", "https://bioconductor.org/biocLite.R"))install.packages("Vennerable", repos="http://R-Forge.R-project.org")

很多包安装失败：

因此自己一个个安装：

if (!requireNamespace("BiocManager", quietly = TRUE))    install.packages("BiocManager")BiocManager::install("")

愣是没搞定，因此sudo R之后继续运行代码：

成功！我安装了三个小时呀！方法错误就是坑爹！

然后继续：

遇到同样的问题，然后作者的意思是版本更新？

于是去看看新的：

cellrouter/CellRouter_Paul_Tutorial.md at master · edroaldo/cellrouter

https://github.com/edroaldo/cellrouter/blob/master/Myeloid_Progenitors/CellRouter_Paul_Tutorial.md

设置工作目录。

缺什么包就安装什么包：

可以运行：

代码的全过程：

在工作目录中，创建一个名为“ results”的文件夹。然后，在文件夹results中，创建一个名为“ paths”的文件夹。

先sudo R以管理员权限进行分析：

#新建文件夹results，然后在results里建立paths文件夹。setwd("/data/yudonglin/Cellrouter/cellrouter/Myeloid_Progenitors")#设置工作目录source('/data/yudonglin/Cellrouter/cellrouter/CellRouter_Class.R')#加载这个CellRouter类，从而能够使用里面的函数library(magrittr) # needs to be run every time you start R and want to use %>%library(dplyr)    # alternatively, this also loads %>%samples 'Paul_Cell_samples.R'))cellrouter 'Paul_Cell_data.R', min.cells=3, min.genes=0)cellrouter 'celltype', metadata.column=cellrouter 'celltype2', metadata.column=## Example regarding how to filter out cells based on a list of variables. This is not used in this tutorial.#mito.genes #percent.mito #cellrouter #cellrouter cellrouter #归一化cellrouter #缩放数据cellrouter #主成分分析build knn graph on the PCA/DC space

plot(cellrouter@pca$sdev, xlab='PC', ylab='Standard deviation of PC')

选择11个主要成分进行tSNE分析。类似于肘部图(碎石图),基于每个主成分对方差解释率的排名。看一个碎石图，看那个点主要在哪个地方没有什么下降趋势。

plotReducedDimension(cellrouter, reduction.type = 'tsne', dims.use = c(1,2), annotation = "celltype", annotation.color = 'celltype_color', showlabels = TRUE, 4.5, 3.5, filename='results/tSNE_1.pdf')

plotReducedDimension(cellrouter, reduction.type = 'pca', dims.use = c(1,2), annotation = "celltype", annotation.color = 'celltype_color', showlabels = FALSE, 4.5, 3.5, filename='results/PCA_1.pdf')

plotReducedDimension(cellrouter, reduction.type = 'tsne', dims.use = c(1,2), annotation = "celltype2", annotation.color = 'celltype2_color', showlabels = TRUE, 4.5, 3.5, filename='results/tSNE_2.pdf')

plotReducedDimension(cellrouter, reduction.type = 'pca', dims.use = c(1,2), annotation = "celltype2", annotation.color = 'celltype2_color', showlabels = FALSE, 4.5, 3.5, filename='results/PCA_2.pdf')

plotReducedDimension(cellrouter, reduction.type = 'dr.custom', dims.use = c(1,2), annotation = "celltype2", annotation.color = 'celltype2_color', showlabels = FALSE, 4.5, 3.5, filename='results/custom_1.pdf')

在xshell运行成功，只是不能直接展示，但是有输出的文件。

### signatures based on sorted populationsmarkers "celltype", pos.only = TRUE, fc.threshold = 

top10 <- markers %>% group_by(population) %>% top_n(10, fc)filename <- 'results/heatmap_top_10_genes_sorted_populations.png'plotSignaturesHeatmap(cellrouter, markers = top10, column.ann = 'celltype', column.color = 'celltype_color', threshold = 2, width = 7, height = 7, filename=filename)

### signatures based on clusters identified by Paul et al.markers <- findSignatures(cellrouter, column = "celltype2", pos.only = TRUE, fc.threshold = 0.5)top10 <- markers %>% group_by(population) %>% top_n(10, fc)filename <- 'results/heatmap_top_10_genes_original_clusters.png'plotSignaturesHeatmap(cellrouter, markers = top10, column.ann = 'celltype2', column.color = 'celltype2_color', num.cells = 100, threshold = 3, width = 15, height = 12, filename=filename)

#Graph-based clustering基于图的聚类cellrouter <- findClusters(cellrouter, method="graph.clustering", num.pcs=15, k=20)plotReducedDimension(cellrouter, reduction.type = 'tsne', annotation="population", annotation.color = 'colors',showlabels = TRUE, width = 4.5, height = 3.5, filename='results/tSNE_graphClustering_clusters.pdf')

#model-based clustering基于模型的聚类cellrouter <- findClusters(cellrouter, method='model.clustering', num.pcs = 15)plotReducedDimension(cellrouter, reduction.type = 'tsne', annotation="population", annotation.color = 'colors',showlabels = TRUE, width = 4.5, height = 3.5, filename='results/tSNE_modelClustering_clusters.pdf')

##Signatures based on clusters identified by CellRoutermarkers <- findSignatures(cellrouter, column = 'population', pos.only = TRUE, fc.threshold = 0.5)filename <- 'results/heatmap_all_genes_CellRouter_clusters.png' #select how many cells to plot as well!plotSignaturesHeatmap(cellrouter, markers, column.ann = 'population', column.color  = 'colors', num.cells=100, threshold =  2.5, width = 15, height = 13, filename = filename)

top10 <- markers %>% group_by(population) %>% top_n(10, fc)filename <- 'results/heatmap_top_10_genes_CellRouter_clusters.png' #select how many cells to plot as well!plotSignaturesHeatmap(cellrouter, top10, column.ann = 'population', column.color  = 'colors', threshold =  2.5, width = 15, height = 13, filename = filename)

#用CellRouter开始轨迹分析libdir '/data/yudonglin/Cellrouter/cellrouter/CellRouter/'#Trajectories can be identified between clusters identified by Paul et al using:cellrouter 10, column.ann = plotKNN(cellrouter, reduction.type = 'tsne', column.ann = 'celltype2', column.color = 'celltype2_color', width = 5, height = 4, filename='results/knn_tsne_original_clusters.pdf')

暂时安装失败。

plotReducedDimension(cellrouter, reduction.type = 'tsne', annotation="celltype2", annotation.color = 'celltype2_color',showlabels = TRUE, width = 4.5, height = 3.5, filename='results/tSNE_original_clusters.pdf')

filename "results/paths/cell_edge_weighted_network.txt"write.table(cellrouter@graph$edges, file=filename, sep='\t', row.names=FALSE, col.names = FALSE, quote=FALSE) #input network#Select population 6. Change sources and targets accordingly, depending on the annotation used (celltype, celltype2 or population, for example)sources '6')targets as.vector(cellrouter@sampTab$population), sources)methods "euclidean", ##Identify trajectoriescellrouter 'population', libdir, paste(getwd(), 

library(igraph)#Preprocess trajectoriesranks 'path_cost', cellrouter 3], num.cells = 

names @pathsinfo##Identify genes regulated along each trajectory (up or down-regulated)cellrouter 'spearman')cellrouter 0.

cellrouter cellrouter 5)save(cellrouter, file='results/CellRouter_Processed.R')

#Create gene regulatory networkgrn.data <- buildGRN(cellrouter, species = 'Mm', genes.use = rownames(cellrouter@ndata), zscore = 5, filename = 'results/GRN.R')#plot clusters of kinetic changes along trajectoriesplotClusterHeatmap(cellrouter, names, 10, 10, 2, 'results/dynamics.pdf')

#Transition from 6.2 in the enrythroid branchplotClusters(cellrouter, names[2], 2, 400, 450, 'results/cluster_dynamics.png')

#Erythroid branch (transition from cluster 6 to 2)cellrouter@top.correlations包含与每个轨迹高度相关或反相关的基因genelist 'Gata2', plottrajectories(cellrouter, '6.2', genelist, rescale = TRUE, columns=1, width=5, height=2, filename='results/dynamics_curve.pdf')

plotPathHeatmap(cellrouter, c('6.2'), genelist, 2, 4, 2, 'results/')

plotDRExpression(cellrouter, genelist, reduction.type='tsne',title="tsne", columns = 4, width=6, height=3, filename='results/tsne_selected_genes_known_Ery.pdf')

#GMP branch (transition from cluster 6 to 7)genelist 'Mpo', plottrajectories(cellrouter, '6.7', genelist, rescale = TRUE, columns=1, width=5, height=2, filename='results/dynamics_curve.pdf')

plotPathHeatmap(cellrouter, c('6.7'), genelist, 2, 4, 2, 'results/')

plotDRExpression(cellrouter, genelist, reduction.type='tsne', columns = 4, width=6, height=3, title="tsne",filename='results/tsne_selected_genes_known_GMP.pdf')

plotbranch(cellrouter, 'up','6.2', '6.7', 1, width=3, height=4, filename='results/6.2_6.7_branch_dynamics.png')

grn.data 'results/GRN.R'))transitions '6.1',grn.scores 'both', flip=               dir.targets='up', columns=2, width=5, height=7, filename='results/GRN_scores')

library('geomnet')p '6.2'plottr(cellrouter, p, grn.scores[[p]]$scores, cluster=FALSE, 2, 4.5, 5.5, paste('results/', p, 'Regulators_6.2_transition.pdf',sep=''))

regulators 'Sphk1',nets1 $GRN,regulators, 

暂时安装不了这个包。

#dynamics of regulators along the transistion from cluster 6 to cluster 7 (GMP branch).p '6.7'plottr(cellrouter, p, grn.scores[[p]]$scores, cluster=FALSE, 2, 4.5, 5.5, paste('results/', p, 'Regulators_6.2_transition.pdf',sep=''))

regulators 'Gfi1', nets2 $GRN,regulators, 

由于这个包安装不了，所以没有结果。

大部分结果都重复出来啦~

参考资料：

master·edroaldo/cellrouter/ CellRouter_Paul_Tutorial.md

https://github.com/edroaldo/cellrouter/blob/master/Myeloid_Progenitors/CellRouter_Paul_Tutorial.md