PCA in seapiper does not use the n_most_varying config parameter

[Nicolai-vKuegelgen]

The PCA in seaPiper can sometimes still look different than the one in the report.
The reason for this seems to be that in the report the n_most_varying paramter from the config is checked and variable genes are trimmed down to that number before calculating PCA (the default in the config file is 500):

mtx <- t(assay(RLD_BLIND))
vars <- sort(apply(mtx, 2, var), decreasing=TRUE)
sel <- names(vars)[vars > 1e-26]
if (config$report$snippet_parameters$Normalisation_QC$n_most_varying > 0)
	sel <- sel[1:min(length(sel), config$report$snippet_parameters$Normalisation_QC$n_most_varying)]
pca <- prcomp(mtx[ , sel], scale.=TRUE)

The reports in seaPiper seem to skip this step, as the PCA there resembles the one created if this subsetting step is omitted:

PCA seapiper	PCA report, default config	PCA report, without config/subsetting