R tryCatch跳过错误

Question

I tried concatenating 2 dataframes: Eset2 and Essential. They share 1 common column which contains gene names and both frames have unique rows.

So I decided to look up the values I need (RMA, ANNOT) in Eset2 and bind them to the row in Essential with corresponding gene name.

But sometimes I wouldn't be able to look up as there are unique genes in Essential. So, there will be an error: my search for a corresponding row in Eset2 would turn up numeric(0).

So I decide to use tryCatch. However, it doesn't help. Instead of putting in NA for RMA and ANNOT when the search found no such gene in Eset2 the script puts in the values from the previous successfully found gene.

    for (i in 1:nrow(essential)) {
      temp <- tryCatch(
      # eset2$GENENAMEand essential[,4] contain gene names
        eset2[eset2$GENENAME == essential[i,4],]$RMA,
      error = function(e) {
        print("This is the 'error' part1")
        return(NA)}
      )
      essential$rma[i] <- temp

      temp <- tryCatch(
        eset2[eset2$GENENAME == essential[i,4],]$ANNOT,
      error = function(e) {
        print("This is the 'error' part2")
        return(NA)}
      )
      essential$long_name[i] <- temp
    }

I solved the issue by using this instead:

        for (i in 1:nrow(essential)) {
            temp <- try(eset2[eset2$SYMBOL == essential[i,4],]$rma)
            if (length(temp) != 0) {
              essential$rma[i] <- temp
            }
            else {
              essential$rma[i] <- NA
            }
            temp <- try(eset2[eset2$SYMBOL == essential[i,4],]$GENENAME)
            if (length(temp) != 0) {
              essential$long_name[i] <- temp
            }
            else {
              essential$long_name[i] <- NA
            }
          }

I wonder if I'm using tryCatch wrong. I tried doing so:

         temp <- eset2[eset2$GENENAME == essential[i,4],]$ANNOT

But it didn't help. Can you see why my tryCatch failed?

Answer 1

I believe the reason the error block is never being evaluated is that a data frame subset evaluating to numeric(0) is not an error. Instead, you should just explicitly check for numeric(0) and replace with NA when you encounter it:

for (i in 1:nrow(essential)) {
    temp <- eset2[eset2$GENENAME == essential[i,4],]$RMA
    essential$rma[i] <- ifelse(length(temp) == 0, NA, temp)

    temp <- eset2[eset2$GENENAME == essential[i,4],]$ANNOT
    essential$long_name[i] <- ifelse(length(temp) == 0, NA, temp)
}

By the way, there is probably a way to vectorize the for loop you used, possibly reducing the entire loop down to a few lines of R code. But this answer hopefully sheds some light on why the error block was never being hit, and what you can do as an alternative.