ProtStatsWF

This package contains workflows for the statistical analysis of proteomics data.

Installation

You can install the development version of ProtStatsWF from GitHub with:

# install.packages("devtools")
devtools::install_github("mpc-bioinformatics/ProtStatsWF")
library(ProtStatsWF)

New features will be introduced in github branches, which are merged as soon as the feature has been properly tested. To install a specific branch of the package please use

# install.packages("devtools")
devtools::install_github("mpc-bioinformatics/ProtStatsWF", ref = "<branchname>")
library(ProtStatsWF)

Workflows

The following workflows are implemented in ProtStatsWF version 0.1.0:

• workflow_QC: Quality control containing normalization, valid values, boxplots, MA-plots, PCA-plots
• workflow_ttest: Statistical comparison of two groups using a t-test (paired or unpaired) together with a volcano plot, histograms of p-values and fold changes as well as boxplots and a heatmap of the biomarker candidates
• workflow_ANOVA: Different versions of ANOVA together with volcano plots histograms of p-values and fold changes as well as boxplots and a heatmap of the biomarker candidates

The following workflows and features are planned for later versions:

• Adding the calculation of on/off proteins to workflow_ttest and workflow_ANOVA
• workflow_clustering: Hierarchical clustering of proteins based on their intensity patterns, visualized by a heatmap and lineplots of individual clusters
• workflow_GO: GO- and Pathway-analysis of a set of biomarker candidates with heatmaps of proteins belonging to a specific term
• workflow_samplesize: Sample size estimation for planning of new experiments

Usage

The data sets have to be present as .xlsx files. The names of columns containing protein intensities have to follow the following scheme: group1_01, group1_02, ...., group2_01, group2_02...

In the following examples the gold standard data set by Uszkoreit et. al, 2022 (PMID: 35845101) is used.

workflow_QC

Normalize the data with the LOESS normalization method:

library(ProtStatsWF)

file <- system.file("extdata", "proteinGroups_D1.xlsx", package = "ProtStatsWF")

workflow_QC(
  data_path = file,
  filetype = "xlsx",
  sep = ",",
  dec = ".",
  header = TRUE,
  sheet = 1,
  output_path = "results",
  output_type = "xlsx",
  intensity_columns = 4:18,
  normalization_method = "loess",
  plot_device = "png",
  PCA_label = TRUE)

workflow_ttest

Use the normalized data to perform a t-test. The data contains 5 groups, we will compare the groups state1 and state2 here:

library(ProtStatsWF)

file <- system.file("extdata", "D1_norm_ID.xlsx", package = "ProtStatsWF")

workflow_ttest(
  data_path = file,
  output_path = "results",
  intensity_columns = 4:9,
  log_before_test = FALSE,
  paired = FALSE,
  column_name_protein = "Gene.names",
  plot_device = "png")

workflow_ANOVA

Use the normalized data to perform an ANOVA. Here we can compare all 5 groups directly:

library(ProtStatsWF)

file <- system.file("extdata", "D1_norm_ID.xlsx", package = "ProtStatsWF")

workflow_ANOVA(
  data_path = file,
  output_path = "results",
  intensity_columns = 4:18,
  paired = FALSE,
  var.equal = FALSE,
  log_before_test = FALSE,
  delog_for_FC = TRUE,
  protein_names_column = "Gene.names",
  plot_device = "png")

Funding

The development and maintanence of ProtStatsWF is funded by CUBiMed.RUB (https://www.cubimed.ruhr-uni-bochum.de/index.html.en). We also also other cool tools and consulting for statistics, bioninformatics and machine learning!