[leopoldguyot]

Implementation of aggregateSamples function. It allows to aggregates the columns of multiples sets of one QFeatures. Methods available for SE or QFeatures.

Related issues: #262 #188 #215

[leopoldguyot]

There is one major issue:
When aggregating several sets at the same time, conflicting coldata data would need to be added to the global colData.
For instance let say we want to aggregate several sets by patients and cell types. For each set the colData columns that are invariants are kept, this means for instance that a run column (this column should contain the same value for all the samples of a set) will be kept. After that when we try to add the new aggregated sets to the QFeatures, the addAssay function will add the information of the colData of the new SummarizedExperiment to the global colData. But if a row is already present in the global colData and that the new set's colData contain clashing data, an error will be thrown. In our example, for one pair of patient/cell type, the run column will contain different values between the different new set (therefore throwing an error).

I can think of different way to fix the problem:

• Only allow the use of the function on a single set (i = character(1))
• Extract the colData of each new Set and remove columns with clashing information before adding them to the QFeatures. (That would add a lot of extra data manipulation)
• Don't add individual colData to the SummarizedExperiment but modify the global colData directly.

The last option means that we cannot simply use the SummarizedExperiment method in the QFeatures method, we need to have a new function that apply the aggregation on a SE but without returning a colData.

What do you think @lgatto ?

[leopoldguyot]

In the new 8.1.0 roxygen release there is now a "support" for setAs.

r-lib/roxygen2#1909 :

roxygen2 didn't recognise setAs() or setGroupGeneric(), so a roxygen block in front of either produced an Rd file with no \usage — which current R-devel reports as Rd files without \usage. setAs() also got no method alias. Both now behave like setMethod() and setGeneric():

\name{coerce,numeric,Account-method}
\alias{coerce,numeric,Account-method}
\usage{
\S4method{coerce}{numeric,Account}(from, to = "Account", strict = TRUE)
}

The problem is that this add this new entry in the QFeatures-class.Rd file:
\S4method{coerce}{MultiAssayExperiment,QFeatures}(from, to = "QFeatures", strict = TRUE)

Which introduces new parameters that were not used in the setAs definition (to and strict). This triggers warning during checks:

❯ checking Rd \usage sections ... WARNING
Undocumented arguments in Rd file 'QFeatures-class.Rd'
‘from’ ‘to’ ‘strict’

Functions with \usage entries need to have the appropriate \alias
entries, and all their arguments documented.
The \usage entries must correspond to syntactically valid R code.
See chapter ‘Writing R documentation files’ in the ‘Writing R
Extensions’ manual.

Should I add these parameters to the setAs method or only document them @lgatto ?