jmacdon / bioc2020anno Goto Github PK

# install annotation packages for species
# BiocManager::install(c("Organism.dplyr", "TxDb.Dmelanogaster.UCSC.dm6.ensGene", "org.Dm.eg.db"))

# load library 
library(Organism.dplyr)

# get annotation
src <- src_organism("TxDb.Dmelanogaster.UCSC.dm6.ensGene")
anno_table <- tbl(src, "id") %>% 
    
    # select desired columns
    dplyr::select(ensembl, symbol) %>% 
    
    # keep one symbol per gene id 
    group_by(ensembl) %>% 
    filter(row_number()==1) %>%
    
    # convert to tibble
    collect()

# add symbols to counts table 
counts_annot <- left_join(counts_tt, anno_table, by = c("feature" = "ensembl"))