| 123456789101112131415161718192021222324 |
- suppressWarnings(suppressMessages(library(clusterProfiler))) # 4.6.0
- suppressWarnings(suppressMessages(library(stringr))) # 1.5.1
- suppressWarnings(suppressMessages(library(data.table))) # 1.14.8
- suppressWarnings(suppressMessages(library(tidyr))) # 1.3.0
- suppressWarnings(suppressMessages(library(dplyr)))
- argv <- commandArgs(TRUE)
- go_file <- argv[1]
- out_file <- argv[2]
- GO_df <- fread(go_file, sep = "\t", header = F)
- colnames(GO_df) <- c("Gene ID","GO")
- GO_df <- GO_df %>%
- separate_rows(GO, sep = ";") %>%
- dplyr::filter(GO != "-")
- GO_df <- as.data.frame(GO_df)
- term2gene <- buildGOmap(GO_df[, c(2, 1)])
- colnames(term2gene)[1] <- "go_id"
- merged_df <- left_join(term2gene, go2term(term2gene$go_id), by = "go_id")
- term2gene <- left_join(merged_df, go2ont(term2gene$go_id), by = "go_id")
- colnames(term2gene) <- c("GO_ID", "Gene", "Term", "Category")
- save(term2gene, file =out_file)
|
