In insightsengineering/hermes: Preprocessing, analyzing, and reporting of RNA-seq data

This vignette explores the design of the show method for HermesData objects.

Idea

The user would like to show some information for a given HermesData object.

Inspiration sources

Default show method

By default we get the following show:

object <- HermesData(summarized_experiment)
show(object)

Compare it with the summary:

summary(object)

DESeq2

Let's see what the DESeq2 package provides.

library(DESeq2)
countData <- matrix(1:100, ncol = 4)
condition <- factor(c("A","A","B","B"))
dds <- DESeqDataSetFromMatrix(countData, DataFrame(condition), ~ condition)
summary(dds)
DESeq2::show(dds)

Seems the same as our default, and less information in fact.

Ideas

Specialized text

We could specialize the text of the show, like in summary(). First check out the show() method code. Then remove some fields and update the titles.

getMethod(f = "show", signature = "SummarizedExperiment")

test_show <- function (object) 
{
    cat("class:", class(object), "\n")
    nms <- assayNames(object)
    if (is.null(nms)) 
        nms <- character(length(assays(object, withDimnames = FALSE)))
    coolcat("assays(%d): %s\n", nms)
    dimnames <- dimnames(object)
    dlen <- sapply(dimnames, length)
    if (dlen[[1]]) 
        coolcat("genes(%d): %s\n", dimnames[[1]])
    else cat("genes: NULL\n")
    coolcat("gene data(%d): %s\n", names(rowData(object, 
        use.names = FALSE)))
    if (dlen[[2]]) 
        coolcat("samples(%d): %s\n", dimnames[[2]])
    else cat("samples: NULL\n")
    coolcat("clinical data(%d): %s\n", names(colData(object)))
}

test_show(object)

rhd <- HermesData(as(summarized_experiment, "RangedSummarizedExperiment"))
test_show(rhd)

Helper Methods to access "additional" genes/samples column names

As we now need this in two places (summary method and show method) it makes sense to factor out the helper methods to access "additional" genes/samples column names so we can also easily access the names of that.

setGeneric("extraColDataNames", function(x, ...) standardGeneric("extraColDataNames"))

setMethod(
  f = "extraColDataNames",
  signature = c(x = "AnyHermesData"),
  definition = function(x, ...) {
    cd_names <- colnames(colData(object))
    cd_req_names <- union(.col_data_non_empty_cols, .col_data_additional_cols)
    setdiff(cd_names, cd_req_names)
  }
)

extraColDataNames(object)

setGeneric("extraRowDataNames", function(x, ...) standardGeneric("extraRowDataNames"))

setMethod(
  f = "extraRowDataNames",
  signature = c(x = "AnyHermesData"),
  definition = function(x, ...) {
    rd_names <- colnames(rowData(object))
    rd_req_names <- union(.row_data_non_empty_cols, .row_data_additional_cols)
    setdiff(rd_names, rd_req_names)
  }
)

extraRowDataNames(object)

Borrowing From summary() show() Method

Start from show() method from summary() and add changes from above.

Note that we can safely assume here: - there are assay names in the object, as at least "counts" needs to be there. - there are row names and column names in the object (also this is forced in validation method).

showMethods("show", classes = "HermesData")
# removeMethod("show", "HermesData")
setMethod(
  f = "show",
  signature = "HermesData",
  definition = function(object) {
    cat_nl(
      "class:",
      S4Vectors::classNameForDisplay(object)
    )
    S4Vectors::coolcat(
      "assays(%d): %s\n", 
      assayNames(object)
    )
    S4Vectors::coolcat(
      "genes(%d): %s\n", 
      rownames(object)
    )
    S4Vectors::coolcat(
      "additional gene information(%d): %s\n", 
      extraRowDataNames(object)
    )
    coolcat(
      "samples(%d): %s\n", 
      colnames(object)
    )
    S4Vectors::coolcat(
      "additional sample information(%d): %s\n", 
      extraColDataNames(object)
    )
  }    
)

Therefore we can use it:

show(object)

Interestingly we can't define this with signature AnyHermesData here, as the inherited SE method takes precedence for some reason.

... and it doesn't interfere with the show() method on HermesDataSummary class.

summary(object)
res <- summary(object)
res
str(res)

During production we need to figure out whether: - it is ok to define for signature AnyHermesData and it works - or if we just define the function separately and then define methods separately for signatures HermesData and RangedHermesData.

Also in production we need to align the summary show method with this, e.g. rename "feature columns" to "gene information", and "sample columns" to "sample information". And reuse the above methods to access these additional names efficiently.

insightsengineering/hermes documentation built on Dec. 15, 2024, 8:07 a.m.