Thread by @Lluis_Rev

My first piece on #multiomics #IBD as part of my thesis is out as a preprint!

I'll summarize it on this thread

medrxiv.org/content/10.110…

https://twitter.com/ibd_bcn/status/1252930880387067905

First, we check how the many parameters of RGCCA #rstats influence and work in their original dataset. We check that different designs select different variables, we discard the superblock design.

2/x

Then we look our own dataset of #IBD samples undergoing a treatment and some controls. We use a method to find good models of the data. The method is available as #rstats package, with its own #pkgdown website llrs.github.io/inteRmodel/
3/x

Then we check this method on another dataset from a previous study involving #pouchitis. On this dataset we get mixed results (see discussion) .
4/x

Then we asses the stability of the models of the #IBD dataset. As you can see on the different iterations of the models they become more accurate.

5/x

Last, we look for the genes and microorganism relevant for each model of the #IBD dataset. We can see on this #UpSetR plots that some of them identify the genes related to location and the microorganisms that separates non-IBD and IBD samples 6/x

I am now validating those genes and microorganisms on other datasets and looking for differences between #IBD and #Colitis #UC.
Hope this provide some direction to prevent or treat #inflammation on the patients and to highlight the role of the #microbiome and other variables 7/7

Links:
📰 Preprint: medrxiv.org/content/10.110…
📦Method/Package: llrs.github.io/inteRmodel/

8/7

Please @threadreaderapp unroll