Plot depth-normalized counts with fitted GAM overlay

For a set of genes, this plots depth-normalized counts across either the coordinate t or the coordinate r, with an optional overlaid fitted function from a mixed GAM object.

Usage

plotGAMestimates(
  Y,
  genes,
  mgam_object,
  curve_fit,
  type = "t",
  nrow = 1,
  include.gam = TRUE
)

Arguments

Y

A numeric matrix-like (e.g., matrix, Matrix::dgCMatrix) of counts with genes in rows and cells/samples in columns. Row names should include the values provided in genes.

genes

A character vector (or integer indices) of gene identifiers to plot.

mgam_object

The output from running the MorphoGAM function.

curve_fit

The output from running the CurveFinder (or interactive version) function.

type

Character string; one of "t" or "r" specifying which coordinate to plot against.

nrow

Integer; number of rows in the facet layout.

include.gam

Logical; if TRUE, overlay the red GAM fit line.

Value

A ggplot2 object showing depth-normalized counts for each gene (facets) versus t or r, with an optional fitted curve.

Details

Columns (cells/samples) are depth-normalized by dividing by the library size (column sum) and multiplying by the median library size across all columns. If type = "t", the x-axis uses curve_fit$xyt$t and fitted values come from mgam_object$fxs.t. If type = "r", the x-axis uses curve_fit$xyt$r and fitted values come from mgam_object$fxs.r. Fitted curves are constructed as \(n_{med} \cdot \exp(\beta_{g0} + f_g(x))\), where \(\beta_{g0}\) is the gene-specific intercept and \(f_g(x)\) is the smooth contribution for that gene.