3: RNA-seq genes to pathways

Author(s) orcid logoMaria Doyle avatar Maria DoyleBelinda Phipson avatar Belinda Phipson
Reviewers Björn Grüning avatarNicola Soranzo avatarHelena Rasche avatar
Overview
Creative Commons License: CC-BY Questions:

  • What are the differentially expressed pathways in the mammary gland of pregnant versus lactating mice?

Objectives:

  • Identification of differentially expressed pathways

Requirements:
Time estimation: 2 hours
Supporting Materials:
Published: Dec 31, 2018
Last modification: Nov 9, 2023
License: Tutorial Content is licensed under Creative Commons Attribution 4.0 International License. The GTN Framework is licensed under MIT
PURL: https://gxy.io/GTN:T00300
Rating: 3.5 (2 recent ratings, 13 all time)
Revision: 13

Sometimes there is quite a long list of genes to interpret after a differential expression analysis, and it is usually infeasible to go through the list one gene at a time trying to understand it’s biological function. A common downstream procedure is gene set testing, which aims to understand which pathways/gene networks the differentially expressed genes are implicated in. There are many different gene set testing methods that can be applied and it can be useful to try several.

The purpose of this tutorial is to demonstrate how to perform gene set testing using tools in Galaxy. The data comes from a Nature Cell Biology paper, EGF-mediated induction of Mcl-1 at the switch to lactation is essential for alveolar cell survival), Fu et al. 2015. That study examined the expression profiles of basal and luminal cells in the mammary gland of virgin, pregnant and lactating mice (see Figure below). How to generate differentially expressed genes from reads (FASTQs) for this dataset is covered in the accompanying tutorials RNA-seq reads to counts and RNA-seq counts to genes. This tutorial is inspired by material from the COMBINE R RNAseq workshop here and the Cancer Research UK workshop here.

Tutorial Dataset. Open image in new tab

Figure 1: Tutorial Dataset
Agenda

In this tutorial, we will deal with:

  1. Import data
  2. Gene Set Testing
    1. Gene Ontology testing with goseq
    2. Gene Set Enrichment Analysis with fgsea
    3. Ensemble gene set enrichment analyses with EGSEA
  3. Conclusion
Comment: Results may vary

Your results may be slightly different from the ones presented in this tutorial due to differing versions of tools, reference data, external databases, or because of stochastic processes in the algorithms.

We will use several files for this analysis:

  • Differentially expressed results files (genes in rows, logFC and P values in columns)
  • Sample information file (sample id, group)
  • Gene lengths file (genes, lengths)
  • Filtered counts file (genes in rows, counts for samples in columns, with lowly expressed genes removed)
  • Gene sets file for MSigDB Hallmark collection for mouse (rdata)

Import data

Hands On: Data upload

  1. Create a new history for this RNA-seq exercise e.g. RNA-seq genes to pathways

    To create a new history simply click the icon at the top of the history panel:

    UI for creating new history

    1. Click on (Edit) next to the history name (which by default is “Unnamed history”)
    2. Type the new name
    3. Click on Save
    4. To cancel renaming, click the “Cancel” button

    If you do not have the (Edit) next to the history name (which can be the case if you are using an older version of Galaxy) do the following:

    1. Click on Unnamed history (or the current name of the history) (Click to rename history) at the top of your history panel
    2. Type the new name
    3. Press Enter

  2. Import the files, there are two options:

    • Option 1: From a shared data library if available (ask your instructor)

      As an alternative to uploading the data from a URL or your computer, the files may also have been made available from a shared data library:

      1. Go into Libraries (left panel)
      2. Navigate to the correct folder as indicated by your instructor.
        • On most Galaxies tutorial data will be provided in a folder named GTN - Material –> Topic Name -> Tutorial Name.
      3. Select the desired files
      4. Click on Add to History near the top and select as Datasets from the dropdown menu

      5. In the pop-up window, choose

        • “Select history”: the history you want to import the data to (or create a new one)
      6. Click on Import

    • Option 2: From Zenodo

      • Copy the link location

      • Click Upload Data at the top of the tool panel

      • Select Paste/Fetch Data

      • Paste the link(s) into the text field

      • Press Start

      • Close the window

    You can paste the names and links below into the Upload tool:

    https://zenodo.org/record/2596382/files/limma-voom_basalpregnant-basallactate
https://zenodo.org/record/2596382/files/limma-voom_luminalpregnant-luminallactate
https://zenodo.org/record/2596382/files/seqdata
https://zenodo.org/record/2596382/files/mouse_hallmark_sets
https://zenodo.org/record/2596382/files/factordata
https://zenodo.org/record/2596382/files/limma-voom_filtered_counts

  3. Add a tag called #basal to the limma-voom_basalpregnant-basallactate and a tag called #luminal to the limma-voom_luminalpregnant-luminallactate files.

    Datasets can be tagged. This simplifies the tracking of datasets across the Galaxy interface. Tags can contain any combination of letters or numbers but cannot contain spaces.

    To tag a dataset:

    1. Click on the dataset to expand it
    2. Click on Add Tags
    3. Add tag text. Tags starting with # will be automatically propagated to the outputs of tools using this dataset (see below).
    4. Press Enter
    5. Check that the tag appears below the dataset name

    Tags beginning with # are special!

    They are called Name tags. The unique feature of these tags is that they propagate: if a dataset is labelled with a name tag, all derivatives (children) of this dataset will automatically inherit this tag (see below). The figure below explains why this is so useful. Consider the following analysis (numbers in parenthesis correspond to dataset numbers in the figure below):

    1. a set of forward and reverse reads (datasets 1 and 2) is mapped against a reference using Bowtie2 generating dataset 3;
    2. dataset 3 is used to calculate read coverage using BedTools Genome Coverage separately for + and - strands. This generates two datasets (4 and 5 for plus and minus, respectively);
    3. datasets 4 and 5 are used as inputs to Macs2 broadCall datasets generating datasets 6 and 8;
    4. datasets 6 and 8 are intersected with coordinates of genes (dataset 9) using BedTools Intersect generating datasets 10 and 11.

    A history without name tags versus history with name tags

    Now consider that this analysis is done without name tags. This is shown on the left side of the figure. It is hard to trace which datasets contain “plus” data versus “minus” data. For example, does dataset 10 contain “plus” data or “minus” data? Probably “minus” but are you sure? In the case of a small history like the one shown here, it is possible to trace this manually but as the size of a history grows it will become very challenging.

    The right side of the figure shows exactly the same analysis, but using name tags. When the analysis was conducted datasets 4 and 5 were tagged with #plus and #minus, respectively. When they were used as inputs to Macs2 resulting datasets 6 and 8 automatically inherited them and so on… As a result it is straightforward to trace both branches (plus and minus) of this analysis.

    More information is in a dedicated #nametag tutorial.

Gene Set Testing

Gene Ontology testing with goseq

We would like to know if there are biological categories that are enriched among the differentially expressed genes. To do this we will perform a Gene Ontology analysis, similar to the RNA-seq ref-based tutorial.

Gene Ontology (GO) analysis is widely used to reduce complexity and highlight biological processes in genome-wide expression studies. However, standard methods give biased results on RNA-seq data due to over-detection of differential expression for long and highly-expressed transcripts.

The goseq tool provides methods for performing GO analysis of RNA-seq data, taking length bias into account. The methods and software used by goseq are equally applicable to other category based tests of RNA-seq data, such as KEGG pathway analysis.

goseq needs 2 files as inputs:

  • a differentially expressed genes file. Information for all genes tested for differential expression (all genes after filtering lowly expressed). This file should have 2 columns:
    • the Gene IDs (unique within the file)
    • True (differentially expressed) or False (not differentially expressed)
  • a gene lengths file. Information to correct for potential length bias in differentially expressed genes. This file should have 2 columns:
    • the Gene IDs (unique within the file)
    • the gene lengths

We need a table of differentially expressed (DE) results for goseq. Here we will use two tables, one for each of the basal and luminal pregnant vs lactate results. This is so we can compare results for the two cell types. These tables were output from the limma-voom tool but DE tables from edgeR or DESeq2 could also be used. For this dataset we will call genes differentially expressed if they have an adjusted P value below 0.01. We can use the gene lengths from the counts table in GEO (provided as a file called seqdata in Zenodo). But if we didn’t have that we could use a tool like featureCounts to output a gene lengths file. The seqdata file contains >20k genes, but we only want the ~15k we have in our differentially expressed genes file. So we will join the lengths file with the differentially expressed genes file, keeping only the length information for genes present in the differentially expressed genes file. We can then cut out the columns we need for the two inputs (gene id, length) (gene id, DE status) and as a bonus they will both be sorted in the same order, which is what we need for goseq.

To generate the two input files we will use:

  • Compute to add a column to the DE tables, that gives genes meeting our adj.P threshold the value “True” and all other genes the value “False”.
  • Join two Datasets to add the gene length information to the differentially expressed genes, matching on gene ids
  • Cut to extract the two columns for the differentially expressed genes information
  • Cut to extract the two columns for the gene length information
Hands On: Prepare the two inputs for GOSeq
  1. Compute an expression on every row with the following parameters:
    • “Add expression”: bool(c8<0.01) (adj.P < 0.01)
    • “as a new column to”: the limma-voom_basalpregnant-basallactate file
    • “Skip a header line”: yes
    • “The new column name”: Status
  2. Join two Datasets side by side on a specified field with the following parameters:
    • “Join”: output of Compute
    • “using column”: Column: 1
    • “with” the seqdata file
    • “and column”: Column: 1
    • “Keep the header lines”: Yes
  3. Cut columns from a table with the following parameters:
    • “Cut columns”: c1,c10 (the gene ids and DE status)
    • “Delimited by”: Tab
    • “From”: output of Join
    • Rename to goseq DE status
  4. Cut columns from a table with the following parameters:
    • “Cut columns”: c1,c12 (the gene ids and lengths)
    • “Delimited by”: Tab
    • “From”: output of Join
    • Rename to goseq gene lengths

We now have the two required input files for goseq.

Hands On: Perform GO analysis
  1. goseq with
    • “Differentially expressed genes file”: goseq DE status
    • “Gene lengths file”: goseq gene lengths
    • “Gene categories”: Get categories
      • “Select a genome to use”: Mouse(mm10)
      • “Select Gene ID format”: Entrez Gene ID
      • “Select one or more categories”: GO: Biological Process
    • “Output Options”
      • “Output Top GO terms plot?” Yes

goseq generates a big table with the following columns for each GO term:

  1. category: GO category
  2. over_rep_pval: p-value for over representation of the term in the differentially expressed genes
  3. under_rep_pval: p-value for under representation of the term in the differentially expressed genes
  4. numDEInCat: number of differentially expressed genes in this category
  5. numInCat: number of genes in this category
  6. term: detail of the term
  7. ontology: MF (Molecular Function - molecular activities of gene products), CC (Cellular Component - where gene products are active), BP (Biological Process - pathways and larger processes made up of the activities of multiple gene products)
  8. p.adjust.over_represented: p-value for over representation of the term in the differentially expressed genes, adjusted for multiple testing with the Benjamini-Hochberg procedure
  9. p.adjust.under_represented: p-value for over representation of the term in the differentially expressed genes, adjusted for multiple testing with the Benjamini-Hochberg procedure

To identify categories significantly enriched/unenriched below some p-value cutoff, it is necessary to use the adjusted p-value.

A plot of the top 10 over-represented GO terms (by p-value) can be output from the goseq tool to help visualise results. Click on the Top over-represented GO terms plot PDF in the history. It should look similar to below.

Basal Plot. Open image in new tab

Figure 2: Basal pregnant vs lactating top 10 GO terms
Question

Can you generate the plot of top 10 GO terms for the luminal differentially expressed genes? Tip: You could use the Rerun button for each step (as the parameters are already selected) and choose the luminal file as input.

Running goseq on the luminal file should give a plot similar to below.

Luminal Plot. Open image in new tab

Figure 3: Luminal pregnant vs lactating top 10 GO terms

For more on GO analysis, including how to simplify GO results and visualize with GO graphs, see the GO Enrichment tutorial.

Gene Set Enrichment Analysis with fgsea

Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) is a widely used method that determines whether a set of genes is enriched in a list of differentially expressed genes. Unlike the previous method with goseq, no threshold is applied for what is considered “differentially expressed”, all genes are used. If a gene set falls at either the top (over-expressed) or bottom (under-expressed) of the list it is said to be enriched. fgsea is a faster implementation of the GSEA method. fgsea requires a ranked list of genes and some gene sets to test.

The Molecular Signatures Database (MSigDB) contains curated collections of gene sets that are commonly used in a GSEA analysis. They can be downloaded from the Broad website. But these collections are only of human gene sets. If working with another species you would need to first map the genes to their human orthologues. However, MSigDB versions for mouse are provided by the Smyth lab here. There are several MSigDB collections, we’ll use the Hallmark collection, which contains 50 gene sets. According to MSigDB, “each gene set in the hallmark collection consists of a “refined” gene set, derived from multiple “founder” sets, that conveys a specific biological state or process and displays coherent expression. The hallmarks effectively summarize most of the relevant information of the original founder sets and, by reducing both variation and redundancy, provide more refined and concise inputs for gene set enrichment analysis”. We’ll use the mouse Hallmark file provided in Zenodo, originally downloaded from http://bioinf.wehi.edu.au/software/MSigDB/mouse_H_v5p2.rdata.

There are several ways we could choose to rank our genes, we could rank by log-fold change (most upregulated to most downregulated) but that doesn’t take into account any error in the log fold change value. Another way is to use the “signed fold change” which is to rank by the sign of the fold change multiplied by the P value (as described here. We could also use the t statistic that’s output from limma, as that takes into account the log-fold change and it’s standard error, see here for more explanation. We’ll use the t statistic to rank here.

Hands On: Perform gene set enrichment with fgsea
  1. Cut columns from a table with
    • “Cut columns”: c1,c6 (the Entrez gene ids and t-statistic)
    • “Delimited by”: Tab
    • “From”: the limma-voom_basalpregnant-basallactate file
  2. Sort data in ascending or descending order with
    • “Sort Query”: the output of Cut
    • “Number of header lines”: 1
    • “Column selections”:
      • “on column”: Column: 2
      • “in”: Descending order
      • “Flavor”: Fast numeric sort (-n)
  3. fgsea with
    • “Ranked Genes”: the output of Sort
    • “File has header?”: Yes
    • “Gene Sets”: mouse_hallmark_sets (this should be rdata format, if not, see how to change it in the Tip below)
    • “Minimum Size of Gene Set”: 15
    • “Output plots”: Yes
  • Click on the pencil icon for the dataset to edit its attributes
  • In the central panel, click Datatypes tab on the top
  • In the Assign Datatype, select your desired datatype from “New type” dropdown
    • Tip: you can start typing the datatype into the field to filter the dropdown menu
  • Click the Save button

fgsea outputs a table of results containing a list of pathways with P values and enrichment scores. It can also output a summary table plot of the top pathways such as the one shown below for the basallpregnant-basallactate contrast below.

fgsea Table. Open image in new tab

Figure 4: fgsea Summary table

An enrichment plot of the each of the top pathways can also be produced, one is shown below. The barcode pattern shows where the genes in the set are found in the list of ranked genes. Most of the bars to the left indicate enrichment of the set at the top of the ranked list of genes (upregulated) and most bars towards the right indicate enrichment at the bottom of the list (downregulated). The enrichment score reflects the degree to which the genes are enriched at the top or bottom of the list.

fgsea Enrichment. Open image in new tab

Figure 5: fgsea Enrichment plot
Question

Can you run fgsea on the luminal contrast and generate the fgsea summary table plot?

Running fgsea on the luminal file should give a plot similar to below.

fgsea Luminal. Open image in new tab

Figure 6: fgsea Summary table for luminal

Ensemble gene set enrichment analyses with EGSEA

The ensemble of genes set enrichment analyses (EGSEA) (Alhamdoosh et al, 2017) is a method developed for RNA-seq data that combines results from multiple algorithms and calculates collective gene set scores, to try to improve the biological relevance of the highest ranked gene sets. EGSEA has built-in gene sets from MSigDB and KEGG for human and mouse. We’ll show here how it can be used with the MSigDB Hallmark collection and KEGG pathways. For input we need a count matrix and EGSEA will perform a limma-voom analysis before gene set testing. We can use the provided filtered counts file output from limma, where the low count genes have been filtered out (output from limma by selecting “Output Filtered Counts Table?”: Yes). We just need to remove the gene symbol and description columns. We also need a symbols mapping file containing just the Entrez ids and symbols, which we can generate from the filtered counts file. The third input we need is a factors information file, containing what groups the samples belong to, we can use the same one from the tutorial RNA-seq counts to genes. EGSEA provides twelve base methods and we will select eleven, all except roast, as the fry method is a fast approximation of roast.

Hands On: Perform ensemble gene set testing with EGSEA
  1. Cut columns from a table (cut) with the following parameters:
    • “File to cut”: limma-voom filtered counts
    • “Operation”: Discard
    • “List of fields”: Select Column:2, Column:3
    • Rename to EGSEA counts
  2. Cut columns from a table (cut) with the following parameters:
    • “File to cut”: limma-voom filtered counts
    • “Operation”: Keep
    • “List of fields”: Select Column:1, Column:2
    • Rename to EGSEA anno
  3. EGSEA with the following parameters:
    • “Count Files or Matrix?”: Single Count Matrix
      • “Count Matrix”: Select EGSEA counts
    • “Input factor information from file?”: Yes
      • “Factor File”: Select factordata
    • “Symbols Mapping file”: EGSEA anno
    • “Contrast of Interest”: basalpregnant-basallactate
    • “Species”: mouse
    • “Gene Set Testing Methods”: Tick camera, safe, gage, zscore, gsva, globaltest, ora, ssgsea, padog, plage, fry
    • “MSigDB Gene Set Collections”: H: hallmark gene sets
    • “KEGG Pathways”: Metabolism and Signalling
    • “Download KEGG pathways?”: Yes
    • “I certify that I am not using this tool for commercial purposes”: Yes

This generates a table of results, and a report. Contained within the report are plots, such as heatmaps of the top ranked pathways, as shown below. Note that we see some similar pathways in the results here as with the fgsea analysis.

EGSEA heatmaps. Open image in new tab

Figure 7: EGSEA heatmaps

KEGG pathway diagrams are generated if KEGG pathways are selected, as shown below. These show the expression values of the genes overlaid, genes upregulated in the contrast are shown in red, downregulated genes in blue. Ribosome was one of the top GO terms identified for the basal pregnant vs lactate contrast and here we see ribosome pathways are in the top ranked KEGG pathways.

EGSEA KEGG. Open image in new tab

Figure 8: EGSEA KEGG pathways
Question

Can you run EGSEA on the luminal contrast identify the top KEGG gene sets? Tip: you could use the rerun button and replace the “Contrast of Interest” with the name of the luminal contrast.

Running EGSEA on the luminal file should give top KEGG gene sets similar to below.

EGSEA KEGG luminal. Open image in new tab

Figure 9: EGSEA KEGG pathways for luminal

Conclusion

In this tutorial we have seen some gene set testing methods that can be used to interpret lists of differentially expressed genes. Multiple methods can be used to help identify pathways of interest and to provide complementary ways of visualising results. This follows on from the accompanying tutorials, RNA-seq reads to counts and RNA-seq counts to genes, that showed how to turn reads (FASTQs) into differentially expressed genes for this dataset. For further reading on analysis of RNA-seq count data and the methods used here, see the articles; RNA-seq analysis is easy as 1-2-3 with limma, Glimma and edgeR (Law et al. 2016) and From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline (Chen, Lun, Smyth 2016).