Datatable error when dividing public repetoir about immunarch

[ahurbanski]

Hey @gustavjoh.

I was facing the same issue trying to use the pubRepApply function. This is how dealt with it:

I searched for the function code with trace(pubRepApply, edit = T) and copied it to my script with another name, but changing the original line

pr.res = dplyr::inner_join(.pr1, .pr2, by = col_before_samples)

for this

pr.res = as.data.table(dplyr::inner_join(.pr1, .pr2, by = col_before_samples))

as it follows:

my_function <- function (.pr1, .pr2, .fun = function(x) log10(x[1])/log10(x[2])) 
{
  col_before_samples = names(.pr1)[1:(match("Samples", colnames(.pr1)) - 
                                        1)]
  tmp = rowMeans(public_matrix(.pr1), na.rm = T)
  .pr1[, (match("Samples", colnames(.pr1)) + 1):ncol(.pr1)] = NULL
  .pr1[["Quant"]] = tmp
  tmp = rowMeans(public_matrix(.pr2), na.rm = T)
  .pr2[, (match("Samples", colnames(.pr2)) + 1):ncol(.pr2)] = NULL
  .pr2[["Quant"]] = tmp
  pr.res = as.data.table(dplyr::inner_join(.pr1, .pr2, by = col_before_samples))
  pr.res[["Samples.x"]] = pr.res[["Samples.x"]] + pr.res[["Samples.y"]]
  pr.res[, `:=`(Samples.y, NULL)]
  names(pr.res)[match("Samples.x", colnames(pr.res))] = "Samples"
  pr.res[["Result"]] = apply(pr.res[, c("Quant.x", "Quant.y")], 
                             1, .fun)
  add_class(pr.res, "immunr_public_repertoire_apply")
}

Then calling the new function:

pr3 = my_function(pr1, pr2)

I don't know how the output was intended to be, but I think it's working fine now.

Hope it helps.

[gustavjoh]

Thank you!

Your solution works, seems like the line causing the problem is the one below.

pr.res[, ':='(Samples.y, NULL)]

What I can see is that it only removes the column "Samples.y" so that line could also be replaced by the code below, perhaps not as clean code. But at least it doesn't change the datatype.

'pr.res = dplyr::select(pr.res, -Samples.y)'

Hope this is fixed in the next version

Thank you again!