Unexpected behavior after recursive call to `subsetByColData`

[cvanderaa]

Hello,

Calling the subsetByColData two times leads to dropping assays when empty assay are generated after the first call. Here is an example:

library(MultiAssayExperiment)
example("MultiAssayExperiment")
(mae <- subsetByColData(mae, mae$age > 40 ))

harmonizing input:
  removing 1 sampleMap rows with 'colname' not in colnames of experiments
A MultiAssayExperiment object of 4 listed
 experiments with user-defined names and respective classes.
 Containing an ExperimentList class object of length 4:
 [1] Affy: SummarizedExperiment with 5 rows and 1 columns
 [2] Methyl450k: matrix with 5 rows and 1 columns
 [3] RNASeqGene: matrix with 5 rows and 1 columns
 [4] GISTIC: RangedSummarizedExperiment with 5 rows and 0 columns

Up to here, everything behaves as expected. The GISTIC assay does not contain a patient with age over 40, therefore it has 0 columns. Now let's reapply the subsetting:

(mae <- subsetByColData(mae, mae$age > 40 ))

A MultiAssayExperiment object of 3 listed
 experiments with user-defined names and respective classes.
 Containing an ExperimentList class object of length 3:
 [1] Affy: SummarizedExperiment with 5 rows and 1 columns
 [2] Methyl450k: matrix with 5 rows and 1 columns
 [3] RNASeqGene: matrix with 5 rows and 1 columns

Here, the GISTIC assay disappeared. Since we apply twice the exact same subsetting, I would expect this output to stay identical to the first output. Note that this observation holds even when the subsetting condition (the logical vector) is different in the second call.

[LiNk-NY]

Hi Chris, @cvanderaa Thanks for pointing this out. I have a patch ready but I will have to do more testing before pushing. See the nodrop branch. Best, Marcel

[LiNk-NY]

Hi Chris, @cvanderaa I've merged the branch and it should be resolved. Thanks! -Marcel

[cvanderaa]

Hi Marcel, @LiNk-NY Thanks a lot for the fix!