Example of BUSCO results to upset plots

busco_to_upset.R

# example of loading busco sample data, preparing gene set data, and making plots

# if necessary, uncomment to install dependencies
# install.packages("readr")
# install.packages("UpSetR")

library(readr)

# function to load BUSCO results and label the first two columns
load_busco <- function(path) {
  df <- read_tsv(path, col_names = FALSE)
  colnames(df) <- c('gene', 'status')
  df
}

tiscar <- load_busco('./data/busco_output/full_table_TiscarSM28_BUSCO')
tokyo <- load_busco('./data/busco_output/full_table_Tokyo1_BUSCO')

# calculate uniqe non-missing genes for each dataset
tiscar_set <- unique(tiscar[tiscar$status != 'Missing',]$gene)
tokyo_set <- unique(tokyo[tokyo$status != 'Missing',]$gene)

# make a list labelling the sets
genesets <- list(
  tiscar = tiscar_set,
  tokyo = tokyo_set
)

library(UpSetR)

# plot
upset(fromList(genesets), nsets=2, order.by="freq")