msigdbr hpc

 install.packages("msigdbr")
 

#BiocManager::install("msigdb")
library(msigdb)
library(ExperimentHub)
library(GSEABase)
 

######1hallmark-----
hallmark_df=msigdbr::msigdbr(species = "Homo sapiens",category = "H") #Mus musculus
hallmark_df
hallmark_list=split(x = hallmark_df$gene_symbol,f = hallmark_df$gs_name)
head(hallmark_list,3)



####2GO-----
GO_df = msigdbr::msigdbr(species = "Homo sapiens",category = "C5") %>% 
  dplyr::select(gene_symbol,gs_exact_source,gs_subcat)
GO_df = GO_df[GO_df$gs_subcat!="HPO",]
GO_df
table(GO_df$gs_subcat)

go_list = split(GO_df$gene_symbol,GO_df$gs_exact_source)
lapply(go_list[1:3], head)



#######3KEGG--------
KEGG_df = msigdbr::msigdbr(species = "Homo sapiens",category = "C2") %>% 
  dplyr::select(gene_symbol,gs_exact_source,gs_subcat)
dim(KEGG_df)
KEGG_df
table(KEGG_df$gs_exact_source) %>%tail()
table(KEGG_df$gs_subcat) %>%tail()


kegg_list = split(KEGG_df$gene_symbol,KEGG_df$gs_exact_source)
lapply(kegg_list[11:13], head)

BiocManager::install("rhdf5filters")
BiocManager::install('GSVA')

library(GSVA)
library(ggplot2)
library(ggpubr)