Skip to contents

This function allows to plot any qualit-control figures available within the shiny app using visualise() or in the HTML report from render().

Usage

# S3 method for nacho
autoplot(
  object,
  x,
  colour = "CartridgeID",
  size = 0.5,
  show_legend = TRUE,
  show_outliers = TRUE,
  outliers_factor = 1,
  outliers_labels = NULL,
  ...
)

Arguments

object

[list] List obtained from load_rcc() or normalise().

x

[character] Character string naming the quality-control metrics to plot from nacho_object. The possible values are:

  • "BD" (Binding Density)

  • "FoV" (Imaging)

  • "PCL" (Positive Control Linearity)

  • "LoD" (Limit of Detection)

  • "Positive" (Positive Controls)

  • "Negative" (Negative Controls)

  • "Housekeeping" (Housekeeping Genes)

  • "PN" (Positive Controls vs. Negative Controls)

  • "ACBD" (Average Counts vs. Binding Density)

  • "ACMC" (Average Counts vs. Median Counts)

  • "PCA12" (Principal Component 1 vs. 2)

  • "PCAi" (Principal Component scree plot)

  • "PCA" (Principal Components planes)

  • "PFNF" (Positive Factor vs. Negative Factor)

  • "HF" (Housekeeping Factor)

  • "NORM" (Normalisation Factor)

colour

[character] Character string of the column in ssheet_csv or more generally in nacho_object$nacho to be used as grouping colour.

size

[numeric] A numeric controlling point size (ggplot2::geom_point() or line size (ggplot2::geom_line()).

show_legend

[logical] Boolean to indicate whether the plot legends should be plotted (TRUE) or not (FALSE). Default is TRUE.

show_outliers

[logical] Boolean to indicate whether the outliers should be highlighted in red (TRUE) or not (FALSE). Default is TRUE.

outliers_factor

[numeric] Size factor for outliers compared to size. Default is 1.

outliers_labels

[character] Character to indicate which column in nacho_object$nacho should be used to be printed as the labels for outliers or not. Default is NULL.

...

Other arguments (Not used).

Examples


data(GSE74821)

autoplot(GSE74821, x = "BD")