PCA

Note

Hiplot website

This page is the tutorial for source code version of the Hiplot PCA plugin. You can also use the Hiplot website to achieve no code ploting. For more information please see the following link:

https://hiplot.cn/basic/pca?lang=en

Principal component analysis (PCA) is a data processing method with “dimension reduction” as the core, replacing multi-index data with a few comprehensive indicators (PCA), and restoring the most essential characteristics of data.

Setup

• System Requirements: Cross-platform (Linux/MacOS/Windows)

• Programming language: R

• Dependent packages: data.table; jsonlite; gmodels, ggpubr, ggplot2

# Install packages
if (!requireNamespace("data.table", quietly = TRUE)) {
  install.packages("data.table")
}
if (!requireNamespace("jsonlite", quietly = TRUE)) {
  install.packages("jsonlite")
}
if (!requireNamespace("gmodels", quietly = TRUE)) {
  install.packages("gmodels")
}
if (!requireNamespace("ggpubr", quietly = TRUE)) {
  install.packages("ggpubr")
}
if (!requireNamespace("ggplot2", quietly = TRUE)) {
  install.packages("ggplot2")
}

# Load packages
library(data.table)
library(jsonlite)
library(gmodels)
library(ggpubr)
library(ggplot2)

Data Preparation

The loaded data are set (gene name and corresponding gene expression value) and sample information (sample name and grouping).

# Load data
data <- data.table::fread(jsonlite::read_json("https://hiplot.cn/ui/basic/pca/data.json")$exampleData[[1]]$textarea[[1]])
data <- as.data.frame(data)
group <- data.table::fread(jsonlite::read_json("https://hiplot.cn/ui/basic/pca/data.json")$exampleData[[1]]$textarea[[2]])
group <- as.data.frame(group)

# Convert data structure
rownames(data) <- data[, 1]
data <- as.matrix(data[, -1])
pca_info <- fast.prcomp(data)
## Create configuration
conf <- list(
  dataArg = list(
    list(list(value = "group")),  # Color by group
    list(list(value = ""))         # No shape group
  ),
  general = list(
    title = "Principal Component Analysis",
    palette = "Set1"
  )
)
## Perform PCA - Note: data must be transposed because PCA analyzes samples (columns)
pca_info <- prcomp(t(data), scale. = TRUE)
## Prepare plot data
axis <- sapply(conf$dataArg[[1]], function(x) x$value)
## Process color grouping
if (is.null(axis[1]) || axis[1] == "") {
  colorBy <- rep('ALL', ncol(data))
} else {
  ## Ensure sample order matches
  colorBy <- group[match(colnames(data), group$sample), axis[1]]
}
colorBy <- factor(colorBy, levels = unique(colorBy))
## Create PCA data frame
pca_data <- data.frame(
  sample = rownames(pca_info$x),
  PC1 = pca_info$x[, 1],
  PC2 = pca_info$x[, 2],
  colorBy = colorBy
)
## Calculate explained variance
variance_explained <- round(pca_info$sdev^2 / sum(pca_info$sdev^2) * 100, 1)

# View data
str(data)

 num [1:9, 1:6] 6.6 5.76 9.56 8.4 8.42 ...
 - attr(*, "dimnames")=List of 2
  ..$ : chr [1:9] "GBP4" "BCAT1" "CMPK2" "STOX2" ...
  ..$ : chr [1:6] "M1" "M2" "M3" "M8" ...

str(group)

'data.frame':   6 obs. of  2 variables:
 $ sample: chr  "M1" "M2" "M3" "M8" ...
 $ group : chr  "G1" "G1" "G1" "G2" ...

head(pca_data)

    sample        PC1         PC2 colorBy
M1      M1  0.8626164  2.17168331      G1
M2      M2  2.1114348  0.50696347      G1
M3      M3  2.9706882 -1.81112892      G1
M8      M8 -3.0779404 -0.85045239      G2
M9      M9 -2.5038211  0.08748266      G2
M10    M10 -0.3629779 -0.10454813      G2

Visualization

# PCA
p <- ggplot(pca_data, aes(x = PC1, y = PC2, color = colorBy)) +
  geom_point(size = 4, alpha = 0.8) +
  geom_hline(yintercept = 0, linetype = "dashed", color = "gray70") +
  geom_vline(xintercept = 0, linetype = "dashed", color = "gray70") +
  stat_ellipse(level = 0.95, show.legend = FALSE) +
  ggtitle(conf$general$title) +
  labs(
    x = paste0("PC1 (", variance_explained[1], "%)"),
    y = paste0("PC2 (", variance_explained[2], "%)"),
    color = axis[1]
  ) +
  
  # Custom color scheme
  scale_color_brewer(palette = conf$general$palette) +
  
  # Add sample labels
  geom_text(aes(label = sample), 
            hjust = 0.5, vjust = -1, size = 3.5, show.legend = FALSE) +
  
  # Theme settings
  theme_bw(base_size = 12) +
  theme(
    plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
    axis.title = element_text(size = 14, face = "bold"),
    axis.text = element_text(size = 12),
    legend.title = element_text(size = 12, face = "bold"),
    legend.text = element_text(size = 11),
    legend.position = "right",
    panel.grid.major = element_line(color = "grey90", linewidth = 0.3),
    panel.grid.minor = element_blank(),
    panel.border = element_rect(fill = NA, color = "grey50", linewidth = 0.5),
    aspect.ratio = 1
  )

# Display plot
p

Too few points to calculate an ellipse
Too few points to calculate an ellipse

Warning: Removed 2 rows containing missing values or values outside the scale range
(`geom_path()`).

PCA

Different colors represent different samples, which can explain the relationship between principal components and original variables. For example, M1 has a greater contribution to PC1, while M8 has a greater negative correlation with PC1.