Need to conditionally apply function to every element in a character matrix

Question

I'm trying to generate a table that only consists of numbers and "" (blank spaces). I need the numbers to be formatted using scientific notation. The issue is, is that if I make this matrix of a numeric type, then I can't use "" since it is not a number, but I need it to be a blank space. If I make this matrix of character type, then the 'scientific' function from the scales library does not recognize the numbers since they are numeric type. If I try to write an if statement within a for loop that iterates through every element in my matrix, then the condition is only applied to the first element generating the following error below:

Error in if (moduleTraitPvalueSig != "") { : missing value where TRUE/FALSE needed In addition: Warning message: In if (moduleTraitPvalueSig != "") { : the condition has length > 1 and only the first element will be used

I am using a matrix called moduleTraitPvalueSig, and the code that geneates the error above is as such

library(scales)
moduleTraitPvalueSig <- moduleTraitPvalue;
moduleTraitPvalueSig[moduleTraitPvalueSig>=0.05]<- ""
for(row in 1:nrow(moduleTraitPvalueSig)) {
  for(col in 1:ncol(moduleTraitPvalueSig)) {
    if(moduleTraitPvalueSig!="")
    {scientific(moduleTraitPvalueSig)}
  }
}

Error in if (moduleTraitPvalueSig != "") { :

missing value where TRUE/FALSE needed

In addition: Warning message:

In if (moduleTraitPvalueSig != "") { :

the condition has length > 1 and only the first element will be used

Answer 1

Don't confuse number use and calculations with rendering. You shouldn't be worrying about scientific notation and spaces until you are ready to print/render.

Some sample data:

m <- matrix(pi, nr=3, nc=3)
m[1,2] <- m[2,3] <- NA
m
#          [,1]     [,2]     [,3]
# [1,] 3.141593       NA 3.141593
# [2,] 3.141593 3.141593       NA
# [3,] 3.141593 3.141593 3.141593

You don't need a loop, just convert all at once.

m[] <- scales::scientific(m)
m
#      [,1]       [,2]       [,3]      
# [1,] "3.14e+00" "NA"       "3.14e+00"
# [2,] "3.14e+00" "3.14e+00" "NA"      
# [3,] "3.14e+00" "3.14e+00" "3.14e+00"

Unfortunately, scale::scientific cares not about NA (odd!), but that doesn't change things much for us, we can re-assign this way:

m[ m == "NA" ] <- ""
m
#      [,1]       [,2]       [,3]      
# [1,] "3.14e+00" ""         "3.14e+00"
# [2,] "3.14e+00" "3.14e+00" ""        
# [3,] "3.14e+00" "3.14e+00" "3.14e+00"

---

Other notes about your code/errors:

• the condition has length > 1 is because if (condition) { ... } requires a condition of length precisely 1. Since you are comparing your whole object, it is returning a length (of TRUE / FALSE / NA ) equal in length to the total dimensions of your object; that is, if your matrix is 3x5, then your if statement is being given 15 logicals. It takes only one, anything else is an error. Vectorizing if/else is done with ifelse(cond, true_expr, false_expr) , but that's a different story (and you do not need it at all here). There are countless questions on SO that reference this precise error message, I suggest you can do a bit more searching on it for more info.

• missing value where TRUE/FALSE needed is often related, given when the length of logical returns is 0 instead of 1 (expected) or more than 1 (the other error you received). This error as well is very frequent on SO, research will give you plenty of examples and ways to investigate/fix it.

• if you are going to do the matrix comparisons element-wise (very inefficient in R given what you intend to do), then at least you can use your for loop values. Using the m matrix I generated above, that would turn the code into something like:

   # you could use this in place of `m[ m == "NA" ] <- ""` for (i in seq_len(nrow(m))) { for (j in seq_len(ncol(m))) { if (m[i,j] == "NA") m[i,j] == "" } } 

  but please don't use it. If you do, don't say I told you to do that kind of replacement in this way, I'll lose what little credibility I've got :-)