What is the "state of the art" in gene set analysis in 2019?

Question

[Istvan Albert]

I am trying to catch up with the latest development in the gene set enrichment analysis and I am looking for recommendations describing practices and approaches that I might have missed. Perhaps older methods are still the best.

Here is the question:

Suppose you ran an analysis pipeline that produced a set of genes or transcripts (for example a set of differentially expressed transcripts across two conditions)

What is the tool or methodology of your choice to help you make sense of the role and function of your set?

[igor]

Based on personal experience, most of the packages are just elaborate wrappers for phyper() (at least in the R universe). So much depends on the actual reference pathways.

Answer 1

[zizigolu]

Actually I like cluster profiler because we can look at a comparative view of pathways or GO terms in experimental conditions

https://yulab-smu.github.io/clusterProfiler-book/chapter12.html

Also I like GOplot circus for nice visualization of DAVID results

https://wencke.github.io/#display-of-the-relationship-between-genes-and-terms-gochord

[Istvan Albert]

The clusterprofiler seems like a quite sophisticated tool - a bit overcomplicated in that it is not clear what data goes into it, so one has devote quite the effort to get their data in the right shape and form - but after that looks like it does quite a bit

https://github.com/YuLab-SMU/clusterProfiler

[zizigolu]

I do agree with you that clusterprofiler does a bit considering demanding efforts to reshaping input data; But I liked that part one can look at pathways and GO terms between conditions comparatively (also defining up and down regulation genes in each condition)

Answer 2

[ATpoint]

I found gprofiler useful as it combines access to various sources such as GO, KEGG, reactome and WP with a GUI to conveniently explore data. Input is a list of gene names.

[Istvan Albert]

They (the Avi Mayan' lab) have also built Enrichr, Clustergrammer, and other tools:

http://amp.pharm.mssm.edu/Enrichr/

https://clustergrammer.readthedocs.io/getting_started.html

[jared.andrews07]

Enrichr is about as easy as it gets in terms of use. I've used both it and clusterProfiler (/ReactomePA/DOSE) quite a bit. clusterProfiler is definitely a bit more annoying to set up, but the visualizations are quite good for more complicated analyses.