Manipulation of limma data for visualisation and comparisons¶

Now we would like to extract the most differentially expressed genes in the various conditions, and then visualize them using an heatmap of the normalized counts for each sample.

Extract the most differentially expressed genes (PRJNA630433 / limma)¶

Basically, we navigate in the limma history of the PRJNA630433 use-case and we repeat a limma run, asking in addition for a file containing the normalised counts, these are in log2 counts per million (logCPM).

Note that, as edgeR, limma returns log2 counts per million (logCPM).

limma settings

Differential Expression Method

→ limma-voom
Apply voom with sample quality weights?

→ No
Count Files or Matrix?

→ Separate Count Files
1: Factor/Name

→ Tissue
1: Factor/1: Group

Note that there will be three Groups (ie factor levels) in this analysis: Dc, Mo and Oc.

→ Oc
Counts file(s)

→ select the data collection icon, then 15: Oc FeatureCounts counts
2: Factor/2: Group

→ Mo
Counts file(s)

→ select the data collection icon, then 10: Mo FeatureCounts counts
3: Factor level (you must click on Insert Group)

→ Dc
Counts file(s)

→ select the data collection icon, then 5: Mo FeatureCounts counts
Use Gene Annotations?

→ No
Input Contrast information from file?

→ No - 1: Constrast

--> `Mo-Dc`

2: Constrast (click Insert Contrast)

→ Oc-Dc
3: Constrast (click Insert Contrast)

→ Oc-Mo
Filter Low Counts

→ No, leave folded
Output options

→ Unfold and select Output Normalised Counts Table?: Yes
Advanced options

→ Put P-Value Adjusted Threshold to 0.1 (to be consistent with DESeq settings)

→ Leave other advanced options unchanged
Run Tool

Note that limma is nicer than edgeR (in Galaxy) and return a single extra dataset limma on data ... and others: Normalised counts. Thus, no need here to do collection manipulations.

Beside, as with edgeR, a limma-voom_normcounts.tsv also show up as a html link in the dataset limma on data 4, data 3, and others: Report, that download directly to your local computer if you click it.

Keep the dataset limma on ... and others: Normalised counts for latter, we will use it for the clustered heatmap.

Generate top lists of limma DE genes¶

Select genes with |log2FC > 2| and p-adj < 0.01 with `Filter data on any column using simple expressions`¶

Filter data on any column... settings

Filter

→ limma on data ... others: DE tables ( this is a collection)
With following condition

→ abs(c2) > 2 and c6 < 0.01 this expression is different from the one used for DESeq2 tables because the column structure is different. Actually this is the same expression that the one used for edgeR.
Number of header lines to skip

→ 1 (these tables have an added header !)
Click the Run Tool button

Rename the "filter on..." collection to limma Top gene lists

Compute a boolean value by row¶

This is to determine whether genes in the lists are up or down-regulated

Compute on rows settings

Input file

→ limma Top gene lists ( collection !)
Input has a header line with column names?

→ Yes
1: Expressions
Add expression

→ c2 > 0 tables
Mode of the operation

→ Append
The new column name

→ Regulation
Avoid scientific notation in any newly computed columns

→ No
Click the Run Tool button

Look at the effect of evaluating the expression c2 > 0 in the new column expression in the output datasets.

Transform `True` and `False` values to `up` and `down`, respectively¶

Column Regex Find And Replace settings

Select cells from

→ Compute on collection 32 (or so)
using column

→ 8 (for limma there are now 8 columns...)
Check

→ click the button Insert Check
Find Regex

→ False
Replacement

→ down
Check

→ click another time the button Insert Check
Find Regex

→ True
Replacement

→ up
Click the Run Tool button

rename the collection Column Regex Find And Replace on collection 44 with limma top gene lists - oriented

Split the lists in `up` and `down` regulated lists¶

This will be performed through 2 successive runs of the tool Select lines that match an expression

Select lines that match an expression settings

Select lines from

→ limma top gene lists - oriented
that

→ matching
the pattern

→ \tup (a tabulation immediately followed by the string up)
Keep header line

→ Yes
Click the Run Tool button

Immediately rename the collection Select on collection... to limma top up-regulated gene lists

Redo exactly the same operation with a single change in the setting of the tool Select lines that match an expression

Select lines that match an expression settings

Select lines from

→ limma top gene lists - oriented
that

→ matching
the pattern

→ \tdown (a tabulation immediately followed by the string down)
Keep header line

→ Yes
Click the Run Tool button

Rename the collection Select on collection... to limma top down-regulated gene lists

keep the last three generated collections for later comparison with edgeR and DESeq2 tools

Plotting an heatmap of the most significantly de-regulated genes¶

For this, we are going to collect and gather all significantly de-regulated genes in any of the 3 conditions, and to intersect (join operation) this list with the rLog normalized count table precedently generated.

Use `Advanced cut` to select the list of deregulated genes in all three comparisons¶

advanced cut settings

File to cut

→ limma Top gene lists (this is a collection)
Operation

→ Keep
Delimited by

→ Tab
Cut by

→ fields
List of Fields

→ Column 1
First line is a header line
Click the Run Tool button

Rename this collection of single column datasets limma top genes names

Next we concatenate the three datasets of the previous collection in a single dataset¶

We do that using the Concatenate multiple datasets tail-to-head while specifying how tool

Concatenate multiple datasets tail-to-head while specifying how settings

What type of data do you wish to concatenate?

→ Single datasets
Concatenate Datasets

→ Click on the collection icon and select limma top genes names
Include dataset names?

→ No
Number of lines to skip at the beginning of each concatenation:

→ 1
Click the Run Tool button

Rename the return single dataset as limma Pooled top genes

Next we extract Uniques gene names from the `Pooled top genes` dataset¶

You probably agree that the same gene may be deregulated in the three pair-wise comparisons which we have performed with DESeq2.

Thus we need to eliminate the redundancy, using the tool Unique occurrences of each record.

Unique occurrences of each record settings

File to scan for unique values

→ limma Pooled top genes
Ignore differences in case when comparing

→ No
Column only contains numeric values

→ No
Advanced Options

→ Leave as Hide Advanced Options
Click the Run Tool button

Add a header the list of unique gene names associated we significant DE in any of the comparisons¶

We do this with the tools Add Header

Add Header settings

List of Column headers (comma delimited, e.g. C1,C2,...)

→ limma_All_DE_genes
Data File (tab-delimted)

→ Unique on data 7x...
Click the Run Tool button

Rename the generated dataset limma_All_DE_genes

Intersection (join operation) between the list of unique gene name associated with DE and the rLog-Normalized counts file.¶

This is the moment when we are going to use the dataset limma on ... and others: Normalised counts and intersect it (join operation) with the list of DE genes in all three condition.

To do this, we are going to use the tool Join two files

Join two files settings

1^st file

→ select limma on ... and others: Normalised counts
Column to use from 1^st file

→ 1
2^nd File

→ limma_All_DE_genes
Column to use from 2^nd file

→ 1
Output lines appearing in

→ Both 1st and 2nd files
First line is a header line

→ Yes
Ignore case

→ No
Value to put in unpaired (empty) fields

→ NA
Click the Run Tool button

Rename the single-element output collection limma Log2CPM Normalized counts of DE genes

Plot a heatmap of the Log2CPM Normalized counts of limma DE genes in all three conditions¶

We do this using the Plot heatmap with high number of rows tool

Plot heatmap with high number of rows settings

Input should have column headers - these will be the columns that are plotted

→ Click the collection icon and select limma Log2CPM Normalized counts of DE genes
Data transformation

→ Plot the data as it is
Enable data clustering

→ Yes
Clustering columns and rows

→ Cluster rows and not columns
Distance method

→ Euclidean
Clustering method

→ Complete
Labeling columns and rows

→ Label columns and not rows
Coloring groups

→ Blue to white to red
Data scaling

→ Scale my data by row
tweak plot height

→ 35
tweak row label size

→ 1
tweak line height

→ 24
Run Tool