做植物是一件比较艰苦的事情，不但资源少，而且有限的资源未必还能用的好，就拿Bioconductor上的注释包来说吧，我在「Bioconductor」不要轻易相信AnnotationHub的物种注释包, 里面就提到拟南芥的物种包用的注释其实一直都没有更新。究其原因，是因为拟南芥的物种包里的注释一直是从TAIR的FTP下载，而我另一篇文章TAIR周期性更新的注释原来不在FTP服务器上也说了，最新的拟南芥注释信息是要在另外的地方进行下载。所以，我写了「Bioconductor」再次提醒，研究植物的不要轻易相信你用的注释包， 让大家尝试用enricher解决问题。
但是生活不能苟且，我好歹在生信圈搬了几年砖，遇到困难不能退缩，于是我决定自己构建一个拟南芥的物种包。代码如下：
library(RSQLite)
library(AnnotationForge)
options(stringsAsFactors = F)
# GENE-GO注释的数据框
# ATH_GO_TERM.txt were create
# by `cat ATH_GO_GOSLIM.txt | cut -f 1,6,8,10 > ATH_GO_TERM.txt`
go_df <- read.table("./ATH_GO_TERM.txt",
sep="\t", header = FALSE,
as.is = TRUE)
go_df$V3 <- ifelse(go_df$V3 == "C", "CC",
ifelse(go_df$V3 == "P", "BP",
ifelse(go_df$V3 == "F", "MF", "")))
# http://www.geneontology.org/page/guide-go-evidence-codes
# select high confidence evidence
go_df <- go_df[! go_df$V4 %in% "IEA",]
colnames(go_df) <- c("GID","GO","ONTOLOGY","EVIDENCE")
# GENE-PUB的数据框
pub_df <- read.table("./Locus_Published_20180330.txt.gz",
sep="\t",
header = TRUE)
## 只选择AT开头的基因
pub_df <- pub_df[grepl(pattern = "^AT\\d", pub_df$name),]
pub_df <- cbind(GID=do.call(rbind,strsplit(pub_df$name, split = "\\."))[,1],
pub_df)
## pubmed_id 不能为空
pub_df <- pub_df[!is.na(pub_df$PMID),]
colnames(pub_df) <- c("GID","GENEID","REFID",
"PMID","PUBYEAR")
# GENE-SYMBOL的注释数据库
symbol_df <- read.table("./gene_aliases_20180330.txt.gz",
sep = "\t",
header = TRUE)
symbol_df <- symbol_df[grepl(pattern = "^AT\\d", symbol_df$name),]
colnames(symbol_df) <- c("GID","SYMBOL","FULL_NAME")
# GENE-FUNCTION
func_df <- read.table("./Araport11_functional_descriptions_20180330.txt.gz",
sep = "\t",
header=TRUE)
func_df <- func_df[grepl(pattern = "^AT\\d", func_df$name),]
func_df <- cbind(GID=do.call(rbind,strsplit(func_df$name, split = "\\."))[,1],
func_df)
colnames(func_df) <- c("GID","TXID","GENE_MODEL_TYPE",
"SHORT_DESCRIPTION",
"CURATOR_SUMMARY",
"COMPUTATIONAL_DESCRIPTION")
## 去重复行
go_df <- go_df[!duplicated(go_df),]
go_df <- go_df[,c(1,2,4)]
pub_df <- pub_df[!duplicated(pub_df),]
symbol_df <- symbol_df[!duplicated(symbol_df),]
func_df <- func_df[!duplicated(func_df),]
# no duplicated row
# all GID should be same type, be aware of factor
file_path <- file.path( getwd())
makeOrgPackage(go=go_df,
pub_info = pub_df,
symbol_info = symbol_df,
function_info = func_df,
version = "0.1",
maintainer = "xuzhougeng ",
author="xuzhogueng ",
outputDir = file_path,
tax_id = "3702",
genus = "At",
species = "tair10",
goTable = "go"
)
#install.packages("./org.Atair10.eg.db", repos = NULL,
# type = "source")
最后会在指定目录下生成"org.Atair10.eg.db", 然后就可以用
install.packages("./org.Atair10.eg.db", repos = NULL,
type = "source")
而且我测试了，能和Y叔的clusterProfiler完美结合
library(org.Atair10.eg.db)
org <- org.Atair10.eg.db
ego_down <-enrichGO(gene = DEG_GENES,
OrgDb = org,
keyType = "GID",
ont = "BP"
)
目前我是自己用为主，如果你们有需要，可以按照如下代码进行安装
# 解决依赖包的问题
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install("org.At.tair.db", version = "3.8")
# 安装我的注释包
install.packages("https://raw.githubusercontent.com/xuzhougeng/org.At.tair.db/master/org.Atair10.eg.db.tgz", repos=NULL, type="source")
出现问题，欢迎在我的GitHubhttps://github.com/xuzhougeng/org.At.tair.db上提出issue
参考资料
- http://bioconductor.org/packages/release/bioc/vignettes/AnnotationForge/inst/doc/MakingNewOrganismPackages.html