# read in file
filename <- "C:/Users/Sonia/Downloads/ID for DEG in QPX (trimmed).txt"
diff_gene_data <- read.delim(filename, stringsAsFactors = FALSE)

# separate genes out by large function. The rest of the script only works with proc_data (biological processes)
cell_data <- subset(diff_gene_data, GO.domain2 == "C")
mol_data <- subset(diff_gene_data, GO.domain2 == "F")
proc_data <- subset(diff_gene_data, GO.domain2 == "P")

# grab genes with multiple Go-slim terms
genes <- unique(proc_data$Gene_ID)
multiple_paths <- vector(length = 0)
for(i in 1:length(genes)){
  sub <-subset(proc_data, Gene_ID == genes[i], select = c(Contig, Gene_ID, GO.slim, GO.domain2))
  if(length(unique(sub$GO.slim)) > 2){
    multiple_paths <- append(multiple_paths, genes[i])
  }
}

# proteins with multiple Go-slim terms: "P87027" "Q8VEH3" "Q5ZMA2" "Q5BJ90" "B2RX14" "Q6NVU2" "P62137" "P83877" "Q6PQD5" "Q61749" "O22607" "Q09472" "Q5DU56"

# grab other information for those genes
mult_proc <- subset(proc_data, Gene_ID == multiple_paths[1], select = c(Contig, Gene_ID, GO.slim, GO.term, GO.term.2, Protein.function))
for(i in 2:length(multiple_paths)){
  mult_proc <- rbind(mult_proc, subset(proc_data, Gene_ID == multiple_paths[i], select = c(Contig, Gene_ID, GO.slim, GO.term, GO.term.2, Protein.function)))
}

# This function keeps giving me an error because it says my data is set up as a list, which can't be used in "write"
write(mult_proc, file = "multiple_Goslim_proteins.txt", sep = "\t")