pcaExperiment: Simple PCA analysis

In JokingHero/ORFik: Open Reading Frames in Genomics

Simple PCA analysis

Description

Detect outlier libraries with PCA analysis. Will output PCA plot of PCA component 1 (x-axis) vs PCA component 2 (y-axis) for each library (colored by library), shape by replicate. Will be extended to allow batch correction in the future.

Usage

pcaExperiment(
  df,
  output.dir = NULL,
  table = countTable(df, "cds", type = "fpkm"),
  title = "PCA analysis by CDS fpkm",
  subtitle = paste("Numer of genes/regions:", nrow(table)),
  plot.ext = ".pdf",
  return.data = FALSE,
  color.by.group = TRUE
)

Arguments

df	an ORFik experiment
output.dir	default NULL, else character path to directory. File saved as "PCAplot_(experiment name)(plot.ext)"
table	data.table, default countTable(df, "cds", type = "fpkm"), a data.table of counts per column (default normalized fpkm values).
title	character, default "CDS fpkm".
subtitle	character, default: paste("Numer of genes:", nrow(table))
plot.ext	character, default: ".pdf". Alternatives: ".png" or ".jpg". Note that in pdf format the complex correlation plots become very slow to load!
return.data	logical, default FALSE. Return data instead of plot
color.by.group	logical, default TRUE. Colors in PCA plot represent unique library groups, if FALSE. Color each sample in seperate color (harder to distinguish for > 10 samples)

Value

ggplot or invisible(NULL) if output.dir is defined or < 3 samples. Returns data.table with PCA analysis if return.data is TRUE.

Examples

df <- ORFik.template.experiment()
# Select only Ribo-seq and RNA-seq
pcaExperiment(df[df$libtype %in% c("RNA", "RFP"),])

JokingHero/ORFik documentation built on Dec. 21, 2024, 12:01 a.m.