Pathway analysis

[hoelzer]

The people always want subsequent pathway analysis of the DEGs.

I used the

• Piano R package

for this sometimes. What I really like is

• WebGestalt

that we can maybe also use locally?

The identified GO terms could then be visualized with

• Revigo

It might be also worth to check current literature for new tools performing GSEA or pathway detection based on DEGs.

[hoelzer]

This sounds like an interesting/simple approach to implement and its also based on DESeq2 output:

The DEGs produced by DESeq2 with an absolute Log2FC>1 and FDR-adjusted p-value<0.05459were used as input to a general GO enrichment analysis [5, 74]. Each term was verified with a460hypergeometric test from the GOstats package (v2.54.0) [21] and the p-values were corrected for461multiple-hypothesis testing employing the Bonferroni method [40]. GO terms with a significant462adjusted p-value of less than 0.05 were reduced to representative non-redundant terms with the use463of REVIGO [71]

https://www.biorxiv.org/content/10.1101/2020.07.28.225581v2.full.pdf

Besides, they also have a methods section about Host signaling pathway enrichment

[MarieLataretu]

piano and Webgestalt works (for some species) with #87

for revigo I couldn't find an R package, but ViSEAGO might be something for us: https://link.springer.com/article/10.1186/s13040-019-0204-1

[hoelzer]

good find @MarieLataretu ! okay, when we can't implement revigo, ViSEAGO looks also nice.

But I think for now it's already great to have piano and Webgestalt added.

Afaik when running revigo in the browser one can download R scripts to reproduce the plots, e.g.:

REVIGO.r.gz

But I never actually used them. Maybe one can extract the scripts needed for the revigo functionalities from there - but I guess that's a lot of work.

[MarieLataretu]

hmm, but it's just the plotting script. the interesting values c("term_ID","description","frequency_%","plot_X","plot_Y","plot_size","log10_p_value","uniqueness","dispensability") are hard coded!

[hoelzer]

Yeah, and calculated somehow via the Web Service. So Yeah, I also dont see that it is worth to invest time to get revigo running. Webgestalt is already pretty neat

MarieLataretu notifications@github.com schrieb am Fr., 2. Okt. 2020, 18:01:

hmm, but it"s just the plotting script. the interesting values c("termID","description","frequency%","plot_X","plot_Y","plot_size","log10_p_value","uniqueness","dispensability") are hard coded!

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/hoelzer-lab/rnaseq/issues/8#issuecomment-702817506, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADN2CZ4PTL5ZE3LCLR6AN2LSIX2OTANCNFSM4JIGFCSA .

[hoelzer]

@MarieLataretu I think ViSEAGO is still something we could implement in the future?