pca_kmeans {RQdeltaCT}

pca_kmeans

Description

This function performs principal component analysis (PCA) together with k means analysis for samples, and generate plot that illustrate spatial arrangement of samples based on the two first components and with assignment to k means clusters. PCA analysis can not deal with missing values, thus all samples with at least one missing value are removed from data before analysis.

Usage

pca_kmeans(
  data,
  do.k.means = TRUE,
  k.clust = 2,
  clust.names = c("Cluster1", "Cluster2"),
  sel.Gene = "all",
  point.size = 4,
  point.shape = c(19, 17),
  alpha = 0.7,
  point.color = c("#66c2a5", "#fc8d62"),
  add.sample.labels = FALSE,
  label.size = 3,
  hjust = 0,
  vjust = -1,
  axis.title.size = 11,
  axis.text.size = 10,
  legend.text.size = 11,
  legend.title.group = "Group",
  legend.title.cluster = "Cluster",
  legend.title.size = 11,
  legend.position = "right",
  plot.title = "",
  plot.title.size = 14,
  save.to.tiff = FALSE,
  dpi = 600,
  width = 15,
  height = 15,
  name.tiff = "pca_and_kmeans"
)

Arguments

data

Object returned from make_Ct_ready() or delta_Ct() functions.

do.k.means

Logical: if TRUE (default), k means analysis will be performed.

k.clust

Integer: number of clusters for k means analysis. Default to 2.

clust.names

Character vector with names of clusters, must be equal to the number of clusters specified in the k.clust parameter.

sel.Gene

Character vector with names of genes to include, or "all" (default) to use all genes.

point.size

Numeric: size of points. Default to 4.

point.shape

Integer: shape of points. If do.k.means = TRUE, the number of provided values must be equal to the number of cluster (k.clust). Default to c(19, 17).

alpha

Numeric: transparency of points, a value between 0 and 1. Default to 0.7.

point.color

Character vector containing colors for compared groups. The number of colors must be equal to the number of groups. Default to c("#66c2a5", "#fc8d62").

add.sample.labels

Logical: if TRUE, points labels (names of samples) will be added. Default to FALSE.

label.size

Numeric: size of points labels (names of samples). Default to 3.

hjust

Numeric: horizontal position of points labels. Default to 0.

vjust

Numeric: vertical position of points labels. Default to -1.

axis.title.size

Integer: font size of axis titles. Default to 11.

axis.text.size

Integer: font size of axis text. Default to 10.

legend.text.size

Integer: font size of legend text. Default to 11.

legend.title.group

Character: title of legend for groups. Default to "Group".

legend.title.cluster

Character: title of legend for k means clusters. Default to "Clusters".

legend.title.size

Integer: font size of legend title. Default to 11.

legend.position

Position of the legend, can be "top", "right" (default), "bottom", "left", or "none" (no legend). See description for legend.position parameter in ggplot2::theme() function.

plot.title

Character: title of plot. Default to "".

plot.title.size

Integer: font size of plot title. Default to 14.

save.to.tiff

Logical: if TRUE, plot will be saved as .tiff file. Default to FALSE.

dpi

Integer: resolution of saved .tiff file. Default to 600.

width

Numeric: width (in cm) of saved .tiff file. Default to 15.

height

Numeric: height (in cm) of saved .tiff file. Default to 15.

name.tiff

character: name of saved .tiff file, without ".tiff" name of extension. Default to "pca_and_kmeans".

Value

A list containing object with plot and, if do.k.means = TRUE, a confusion matrix that show classification performance of k means method. Created plot is also displayed on the graphic device.

Examples

library(tidyverse)
data(data.Ct)
data.CtF <- filter_Ct(data.Ct,
                      remove.Gene = c("Gene2","Gene5","Gene6","Gene9","Gene11"),
                      remove.Sample = c("Control08","Control16","Control22"))
data.CtF.ready <- make_Ct_ready(data.CtF, imput.by.mean.within.groups = TRUE)
data.dCt <- delta_Ct(data.CtF.ready, ref = "Gene8")
pca_kmeans(data.dCt, sel.Gene = c("Gene1","Gene16","Gene19","Gene20"))