data-raw/freq_go_pairs.R

In Berghopper/GAPGOM: GAPGOM (novel Gene Annotation Prediction and other GO Metrics)

# In this file internal procedures will be defined for generating data, this is for "developers only".

library(data.table)
options(digits=22, stringsAsFactors = FALSE)

# generate score matrix between most frequent occuring gos in all organisms/ontologies
.gen_semscor_topo_matrices <- function(old_listy = NULL, amount=2) {
  # old_listy is deprecated
  options(digits=22, stringsAsFactors = FALSE)
  ids <- c("ENTREZ", "ENSEMBL")
  organisms <- c("mouse", "human")
  ontologies <- c("MF", "CC", "BP")
  combos <- expand.grid(ids, ontologies, organisms, stringsAsFactors = FALSE)
  my_names <- apply(combos, 1, function(row) {paste0(row, collapse = "_")})
  #colnames(combos) <- c("id", "organism", "ontology")
  res <- lapply(seq_len(nrow(combos)), function(i) {
    row <- combos[i,]
    id <- row[[1]]
    ontology <- row[[2]]
    organism <- row[[3]]
    tmp_godat <- GAPGOM::set_go_data(organism, ontology, computeIC = TRUE)
    geneanno <- data.table(tmp_godat@geneAnno)
    counted <- geneanno[unique(geneanno[[1]]), .N, by=GO, on=.(ENTREZID)]
    top_gos <- counted[rev(order(counted$N)),][1:amount,][[1]]
    # now that we have the top gos, generate the matrix.
    custom <- setNames(as.list(top_gos), seq(top_gos))
    resmat <- GAPGOM::topo_ic_sim_genes(organism, ontology, c(), c(), 
                                        custom_genes1 = custom, 
                                        custom_genes2 = custom, 
                                        go_data = tmp_godat, 
                                        use_precalculation = FALSE, 
                                        garbage_collection = TRUE, idtype = id)$AllGoPairs
    save(resmat, file=paste0("./", my_names[i] ,".RData"), compress = "xz", compression_level = 9)
    return(resmat)
  })
  names(res) <- my_names
  return(res)
}
res <- .gen_semscor_topo_matrices(amount=350)
# add sessioninfo to stored result, very important to see if the matrix is still
# up-to-date.
res$sessioninfo <- sessionInfo()
save(res, file="./res.RData", compress = "xz", compression_level = 9)