Indentify highly variable genes — filt_gex

Identifies highly variable features from a log-normalized count matrix or filters matrices by a list of genes provided by the user.

Usage

filt_gex_byhvg(pb_dat_list, prior_hvg = NULL, var.threshold = 1)

Arguments

pb_dat_list: List of SummarizedExperiment generated from MOFAcellulaR::filt_profiles()
prior_hvg: NULL by default. Alternatively, a named list with a character vector containing features to select.
var.threshold: Numeric. Inherited from scran::getTopHVGs(). Minimum threshold on the metric of variation

Value

A named list of SummarizedExperiments per cell type provided with filtered normalized log transformed data in their logcounts assay

Details

This function estimates highly variable genes per cell type using scran::getTopHVGs. Alternatively, this function allows the user to provide the features to be used in each cell type. If prior genes are used, for cell types where this information is missing, highly variable genes will be calculated

Examples

inputs_dir <- base::system.file("extdata", package = "MOFAcellulaR")
load(file.path(inputs_dir, "testpbcounts.rda"))
load(file.path(inputs_dir, "testcoldata.rda"))

pb_obj <- create_init_exp(counts = testpbcounts,
                          coldata = testcoldata)

ct_list <- filt_profiles(pb_dat = pb_obj,
                         cts = c("Fib","CM"),
                         ncells = 5,
                         counts_col = "cell_counts",
                         ct_col = "cell_type")

ct_list <- filt_gex_byexpr(pb_dat_list = ct_list,
                           min.count = 5,
                           min.prop = 0.25)
#> Warning: All samples appear to belong to the same group.
#> Warning: All samples appear to belong to the same group.

ct_list <- tmm_trns(pb_dat_list = ct_list,
                    scale_factor = 1000000)

ct_list <- filt_gex_byhvg(pb_dat_list = ct_list,
                          prior_hvg = NULL,
                          var.threshold = 0)