批量安装一系列github上面的R包

Posted on by

最近在我的授课电脑上面一股脑更新了全部的R套件，包括R和rstudio，以及rtools，还有几百个r包文件夹都删除了。

但是在参考我两年前的教程：一口气安装800个R包，批量安装800多个R包的时候，发现下面的这些包其实是在github上面：

ConsensusTME DoubletFinder EPIC MCPcounter SCENIC SCEVAN SCopeLoomR VISION copykat cpbnplot depmap ggforestplot immunedeconv jjAnno mMCPcounter presto scMetabolism spatstat.core tgutil xCell yaGST

本来应该是使用 devtools::install_github(‘satijalab/seurat-data’) 这样的代码进行安装，但是我因为电脑在长城内部，所以没办法访问github，会得到如下所示的报错：

> devtools::install_github('satijalab/seurat-data')
Downloading GitHub repo satijalab/seurat-data@HEAD
Error: Failed to install 'SeuratData' from GitHub:
 error reading from connection

所以就查询了一下他们的github链接，也是很容易构造，一个简单的案例是：

https://api.github.com/repos/satijalab/seurat-data/tarball/HEAD

最后全部的列表如下所示：

https://api.github.com/repos/cansysbio/ConsensusTME/tarball/HEAD

https://api.github.com/repos/chris-mcginnis-ucsf/DoubletFinder/tarball/HEAD


https://api.github.com/repos/GfellerLab/EPIC/tarball/HEAD


https://api.github.com/repos/ebecht/MCPcounter/tarball/master


https://api.github.com/repos/aertslab/SCENIC/tarball/HEAD


https://api.github.com/repos/AntonioDeFalco/SCEVAN/tarball/HEAD


https://api.github.com/repos/aertslab/SCopeLoomR/tarball/HEAD


https://api.github.com/repos/YosefLab/VISION/tarball/HEAD


https://api.github.com/repos/navinlabcode/copykat/tarball/HEAD


https://api.github.com/repos/noriakis/cpbnplot/tarball/HEAD


https://api.github.com/repos/nightingalehealth/ggforestplot/tarball/HEAD


https://api.github.com/repos/omnideconv/immunedeconv/tarball/HEAD


https://api.github.com/repos/junjunlab/jjAnno/tarball/HEAD


https://api.github.com/repos/cit-bioinfo/mMCP-counter/tarball/HEAD


https://api.github.com/repos/immunogenomics/presto/tarball/HEAD


https://api.github.com/repos/wu-yc/scMetabolism/tarball/HEAD


https://api.github.com/repos/spatstat/spatstat.core/tarball/HEAD


https://api.github.com/repos/tanaylab/tgutil/tarball/HEAD


https://api.github.com/repos/dviraran/xCell/tarball/HEAD

上面的链接每个背后都是一个文件，需要自己手动点击下载，得到如下所示的文件：


$ ls -lht |cut -d" " -f 5-

86M Sep 19 10:19 omnideconv-immunedeconv-v2.1.2-0-g0c5c619.tar.gz 
 19M Sep 19 10:19 dviraran-xCell-1.3-2-g7339410.tar.gz
1.3M Sep 19 10:19 spatstat-spatstat.core-v2.4-3-19-g6c80ceb.tar.gz
 28K Sep 19 10:19 tanaylab-tgutil-0.1.4-61-g0e4a2e8.tar.gz
770K Sep 19 10:19 wu-yc-scMetabolism-4f6f8f1.tar.gz
663K Sep 19 10:19 immunogenomics-presto-31dc97f.tar.gz
 19K Sep 19 10:19 cit-bioinfo-mMCP-counter-v1.1.0-3-gcec6971.tar.gz
1.5M Sep 19 10:19 junjunlab-jjAnno-0.0.3-13-ge8de4ce.tar.gz
 15M Sep 19 10:19 NightingaleHealth-ggforestplot-547617e.tar.gz
6.9M Sep 19 10:19 navinlabcode-copykat-b795ff7.tar.gz
574K Sep 19 10:19 noriakis-cpbnplot-f279229.tar.gz
 27M Sep 19 10:19 YosefLab-VISION-v3.0.1-0-g8dc5c4e.tar.gz
6.6M Sep 19 10:19 AntonioDeFalco-SCEVAN-1.0.0-32-g8048b45.tar.gz
 15M Sep 19 10:19 aertslab-SCENIC-v1.3.0-1-g06d3f34.tar.gz
587K Sep 19 10:19 aertslab-SCopeLoomR-v0.3.1-83-g20f4e0a.tar.gz
4.5M Sep 19 10:19 GfellerLab-EPIC-v1.1.7-0-g50a4f40.tar.gz
 93K Sep 19 10:19 ebecht-MCPcounter-b6eac73.tar.gz
1.4M Sep 19 10:19 chris-mcginnis-ucsf-DoubletFinder-1b1d4e2.tar.gz
766K Sep 19 10:19 cansysbio-ConsensusTME-b3c8d3a.tar.gz
 19M Sep 18 11:48 genecell-COSGR-v0.9.0-1-gc8f3f53.tar.gz
 36M Sep 18 10:27 zhanghao-njmu-SCP-v0.5.1-0-g458ce17.tar.gz

这样的github包的安装方法也是有两个：

## 使用devtools::install_local自动安装SCP及相关依赖（这种安装过程可以自动寻找安装依赖包）：
devtools::install_local("SCP-0.5.1.tar.gz")
### 或使用install.packages安装SCP（如果缺失依赖包会报错，需要再根据提示手动逐一安装依赖包）：
install.packages("SCP-0.5.1.tar.gz", repos = NULL, type = "source")

这里，我们建议是使用 devtools::install_local 安装这一系列包，但是我批量安装的代码没有写好：

fs=list.files(pattern = '*tar.gz')
fs
lapply(fs,function(i){
 #i=fs[17]
 print(i)
 if(i %in% installed.packages()[,"Package"] ){
 print('ok')
 }else{
 print('error')
 #devtools::install_local(i, upgrade = c( "never"))
 }
})

因为上面的文件名字太长了，而且文件名的规则有点乱，需要从里面切割出来我们的包的名字，有点难度哦！但是之前看到的另外一系列本地包就很容易切割，详见：听说你无法下载SeuratData的单细胞示例数据 ：

fs=list.files(pattern = '*tar.gz')
fs
library(devtools)
library(stringr)
lapply(fs,function(i){
 #i=fs[17]
 print(i)
 if(strsplit(i,'_')[[1]][1] %in% installed.packages()[,"Package"] ){
 print('ok')
 }else{
 print('error')
 devtools::install_local(i, upgrade = c( "never"))
 }
})
​

Comments are closed.