Network analysis with freeCount | R bloggers

Overview

Weighted gene coexpression network analysis (WGCNA) is used to investigate the function of genes at the system level. In a network analysis, genes with similar expression patterns are grouped into modules. The sets of genes in these modules are expressed as a result of shared biological functions, pathways, tissues, traits, etc.

Which genes share expression patterns between samples?

The free phone The NA app helps you perform network analyzes of normalized gene counts, which can be produced with differential expression analysis tools such as freeCount DA.

Learning objectives

• Learn how to create lists of co-expressed genes in network modules
• Practice adjusting settings to construct gene co-expression networks
• Understand how to identify sets of genes that share system-level functions

Related

This tutorial is the third in a series and uses the TMM normalized data created in the Creating DE gene lists with freeCount self-study.

WGCNA

The construction of co-expression networks using the WGCNA R package is a tricky process, but conceptually simple (DOI: 10.2202/1544-6115.1128). In a co-expression network, the nodes represent genes. The nodes are connected if the corresponding genes are significantly co-expressed in appropriately chosen samples.

Given the assumptions of WGCNA, it is important to properly design your study for network analysis so that you can draw reasonable conclusions from the results. First, WGCNA assumes that the count data has been pre-processed and normalized (DOI: 10.1186/1471-2105-9-559). It is also important to consider whether you have enough samples to build an informative network, in which the co-expression signal is not affected by any given sample.

Before you start

The exercise in this tutorial uses the freeCount apps in RStudio on a PC. Make sure you have downloaded, installed, and installed the following tools up to date on your PC:

1. R software environment
2. RStudio desktop application

For Windows usersadditionally install RTools.

It not possible to use the freeCount NA app online via the free subscription from Posit Cloud, because it requires too much memory.

Input data

1. Download the tribolium normalized counts file
2. Download the tribolium experimental design file

Tip! Right click and select Save as… to download the above files in the necessary formats.

The Analysis App

The following steps show you how to download and use the freeCount Network Analysis (NA) app.

1. Download the free Count R Shiny applications
   1. Go to https://github.com/ElizabethBrooks/freeCount
   2. Click on the green < > Code knob
   3. Click Download ZIP
2. Grab the freeCount-main folder
3. Navigate to the apps folder
4. Open the NO file in RStudio
5. Click Install on the yellow banner to install the necessary R packages (or run the code on lines 10 through 20)
6. Click on the Run app button in the top right corner of the source panel

Analysis process

Complete the following steps to create lists of co-expressed genes in network modules.

1. Upload the data and click Upload
2. Click on the Perform analysis button that appears on the left side of the screen
3. Check the data settings on the Cleaning data tab
4. Adjust the network settings on the Network construction tab
5. Create an informational network and curated list of co-expressed genes by repeating steps 3 and 4
6. Download lists of genes or module eigengenes from the Results tab

1. Upload data

Upload the data and click Upload.

Input data

1. The first file you need to upload is the gene count table with the normalized gene counts for your experiment. In this tutorial we will use the tribolium normalized counts file.
2. The second file you need to upload is the table showing the experimental design describing the samples in your study. In this tutorial we will use the tribolium experimental design file.

2. Perform analysis

Click on the Perform analysis button that appears on the left side of the screen.

3. Check the data settings

Check the data settings on it Cleaning data tab.

The Minimum cluster size of branches And Cutting heights of branches can be customized to help identify and remove outliers from the input data. After changing these settings, see the following Example clustering to detect outliers plot to see which samples do not cluster well with their groups and may need to be removed. The red line is the cutting height used to remove outliers.

4. Adjust the network settings

Adjust the additional settings on the Network construction tab.

Change the Soft threshold force to shift the range of suggested soft thresholds (red numbers) in the following graphs. Soft thresholding assigns a linkage weight to each gene pair.

Then set the Soft threshold force by looking at the above Scale independence plot to see where the recommended scale-free topology model fit falls (red line on the y-axis). Note which red number is closest to the red line on the y-axis. Then take a look at the Mean connectivity plot to see what that number corresponds to in the average connectivity (y-axis).

Then you can use the Module Own Gene Cutting Height in the Network construction section. This allows you to adjust the size of your modules by merging modules based on co-expression similarity.

5. Create composite results

Create an informational network and a curated list of co-expressed genes by repeating steps 3 and 4.

It may be necessary to repeatedly adjust the settings and inspect the network to create a well-curated list of co-expressed genes, grouped into a manageable set of modules.

6. Download results

Download lists of genes or module eigengenes from the Results tab.

The Gene module data table contains the list of co-expressed genes associated with the different network modules. This file can be input into the functional analysis app freeCount FA to investigate the potential functions of the gene sets in each module.

The Intrinsic expression data table contains the eigengene expression data of the network modules. This file can be used in various downstream analyses, such as differential eigengene expression analysis.

Related