Thesis by Antoine Bichat

[abichat]

Full name

Dr Antoine Bichat

Link to GitHub repo

https://github.com/abichat/thesis
https://abichat.github.io/thesis/

Template

https://github.com/abichat/hadamardown

Completion Date

09/12/2020

Institution

Université Paris-Saclay

Summarize your thesis

Discovering multi-scale metagenomic signatures through hierarchical organization of species

This thesis deals with the use of hierarchical information in differential abundance analyses in metagenomics. Taxa that make up the microbiome are usually associated with a tree, like the taxonomy or the phylogeny, that reflects a biological link between them. It is therefore natural to exploit this hierarchical information to increase the statistical power of differential abundance techniques. We first investigated the efficiency of existing hierarchical differential abundance detection procedures and the impact of tree choice on those. We then developped our own hierarchical differentially abundance detection procedure. It models the taxa associated z-scores as realization of an Ornstein-Uhlenbeck process on a tree with shifts on its optimal value then a lasso-like regression is used to identify optimal positions and intensities of the shifts.

What did you like about using {thesisdown} for this work?

• the possibility to use R and markdown to write the manuscript
• the HTML output in addition to the usual pdf one
• the ease to customization
• the collection of templates in the readme
• the reactivity of @ismayc

[ismayc]

Thank you! I'm still thinking about how best to display and acknowledge these, but I'll ping you when yours is ready.