In MassDynamics/lfq_processing: Processes Proteomics Files for Quantification.

knitr::opts_chunk$set(fig.cap = NULL, fig.path = params$output_figure)

library(knitr)
library(data.table)
library(ggplot2)
library(ggrepel)
library(GGally)
library(umap)
library(FactoMineR)
library(factoextra)
library(corrplot)
library(viridis)
library(ggpubr)
library(Hmisc)
library(plotly)
library(stringr)
library(bit64)

#assert that all the stuff we need is there. 
stopifnot(exists("expdes"))
stopifnot(exists("prot"))
stopifnot(exists("prot_int"))

expdes <- expdes[,c("condition", "experiment", "reporter_channel", "replicate")]

pca_dt <- prot_int[, .(
  id,
  condition,
  replicate,
  log2NIntNorm,
  run_id
)]

pca_dt <- dcast(pca_dt, condition + replicate ~ id, value.var = "log2NIntNorm")



num_dimensions = min(pca_dt[, .N-1], 10)
res.pca <- PCA(pca_dt[, 3:ncol(pca_dt)], graph = FALSE, ncp = num_dimensions)
eig.val <- get_eigenvalue(res.pca)
eig.val <- data.table(dims = rownames(eig.val), eig.val)

scree_plot <- ggplot(eig.val[1:num_dimensions], aes(x=reorder(dims, -`variance.percent`), y=`variance.percent`)) +
  geom_bar(stat="identity", fill = "skyblue2") +
  theme_minimal() +
  geom_text_repel(aes(label=(round(`variance.percent`,1))), direction = 'y') +
  scale_x_discrete("PCA components") +
  scale_y_continuous("% Variance") +
  ggtitle("Proteins - PCA Screeplot")


ggplotly(scree_plot, tooltip = c("y")) %>% config(displayModeBar = T, 
                                           modeBarButtons = list(list('toImage')),
                                           displaylogo = F)