library("dplyr")
library("tidyr")

read_msigdb <- function(file) {
  msigdb_cols     = count.fields(file, sep = "\t")
  msigdb_cols_max = max(msigdb_cols[!is.na(msigdb_cols)])
  msigdb_data     = read.table(file, header = FALSE, sep = "\t", quote = "",
                               fill = TRUE, col.names = paste0("V", seq_len(msigdb_cols_max)),
                               stringsAsFactors = FALSE)

  msigdb_data     = tbl_df(msigdb_data) %>%
    select(-V2) %>%
    gather(columns, gene, -V1) %>%
    select(-columns) %>%
    mutate(gene = ifelse(gene == "", NA, gene)) %>%
    filter(!is.na(gene)) %>%
    rename(pathway = V1) %>%
    rename(gene_entrez = gene) %>%
    mutate(gene_entrez = as.character(gene_entrez)) %>%
    arrange(pathway)
  return(msigdb_data)
}

msigdb_data <- read_msigdb(msigdb_file)