Doctoral research in Molecular Genetics with a heavy tint of Developmental Biology. My thesis is called “Importance du contexte cellulaire et de la régulation spatio-temporelle de l’expression du facteur de transcription Otx2 dans la modulation de ses fonctions”, but please don’t go read it. Our lab was focused on the various roles of the gene OTX2. Otx2 is a fascinating, fascinating gene because it belongs in kind of the same family as the Hox genes; it’s a transcription factor whose function is dramatically important during development (it literally makes sure you have a head), and also reappears later on in life with similarly important (although more bodily restricted) roles in the retina and in the brian in particular.

Trying to tease out its various functions is exactly what large-scale analyses were made for, and therefore that’s what we used ! We setup a ChIP-seq and interactome model in the adult retina to investigate how exactly Otx2 accomplishes its function. I’ll copy-paste from myself and say that cross-referencing ChIP-seq and interactome data, we showed that while Otx2 shares many binding sites with its homologue Crx, it forms completely different protein complexes, indicating different functional profiles for the various members of the Otx family.

My favorite part of this thesis, however, is the one I didn’t get time to publish at the time, and will probably never see published ever by now. It dealt with the notions of heterochrony and neoteny as outlined by Stephen Jay Gould n his 1977 book Ontogeny and Phylogeny. Essentially, genes such as Otx2 pop up at some point of embryonic development, do their thing for a little while, and disappear, leaving you with a new limb, a new head, a new torso - they’re that powerful. So… What would happen if any of them happened to be expressed just that little longer ? Would you get an extra limb ? Two torsos ? A bigger head ? Dr Gould says yes to the last one. We wanted to check if that was true in at least some way. We developed a nifty inducible system where you could turn Otx2 on and off on command and tried to play with its time window of expression. The system took forever to set up, was very imperfect, and yielded completely unexpected results - taken together, this produced an article that I still love to this day, but didn’t make it through reviews in time before I graduated, and is still floating in limbo ever since.

My mentor was Thomas Lamonerie, who taught me both that intellectual curiosity should apply above all else to any question, as well as how to trim down experiments and reasonings to their leanest core.

Master’s studies in biosciences at one of France’s top research and teaching institutions. Extensive training in:

• Molecular Biology
• Genetics (Molecular Genetics and Population Genetics)
• Epigenetics
• Developmental Biology (animal and vegetal, as well as evo-devo)
• Ecology

I was lucky enough to benefit from the structure of the Masters then, where multiple internship periods were offered, giving you the chane to explore aspects of biology you wouldn’t necessarily be able to make the time for otherwise. As such, I was able to help evaluating the link between sleep and hippocampus function by setting up a classic memory swim test for rats in a new lab. I also modeled the link between food abundance and the reproductive success of the Great Owled Horn as a model for conservation strategies. Fun fact, the only EU-subsidized conservation strategy for raptors was to release rabbits in the territory of threatened species. The reasoning was to give the raptors easy food so they can stop worrying about it and spend more time reproducing, or at least live longer. Well, we showed that’s not the case. Raptors generally have a close to 100% kill rate, so will pretty much do great regardless unless their prey is endangered itself. Which rabbits are not.

I used my training gathered here during the other Master in a final internship modeling axin flows in A. thaliana meristem in the Laboratoire Reproduction et Developpement des Plantes. It wasn’t a great success, and the model that I built was horribly inefficient, but it worked, and it was my first foray into the use of data and models in Biology. I learned a lot under the direction of Olivier Hamant and Pradeep Das, and through interactions with my friend Benoit Landrein who did his Masters with me there, got his PhD there, and is now having a brilliant carrer there.

Master’s studies in complex systems at one of France’s top research institutions. This was a time where Data Science was not really something thought of in biology - The first sequenced genome was sequenced 5 years earlier, microarray itself was still a kinda new thing… So it was hard to find specific training on applying big data techniques to biology. Inasmuch as people almost don’t use it anymore these days, I don’t even think the term “big data” existed yet then. So when I started looking at opportunities to extend my knowledge in the domain, this (at the time) brand new Master appeared as an evidence.

It was all built around the idea of modeling and of extracting meaningful information from very complex systems, be they complex by their number of rules or the immense of data they generate, or both. Topics included:

• Emergent behavior and single-agent modeling
• Error propagation
• Differential equations, especially their chaotic behavior
• Constraint graph models
• Probabilistic modeling

I feel incredibly lucky to have been able to follow this course, as I learned so much in it when it comes to way to approach a complex, data-heavy question.

• Python
  • overwhelmingly my favorite language. Unoriginal, I know, but I like the modularity, the accessibility, and most of all the ecosystem
• R
  • because I don’t think there exists a bioinformatician without R knowledge. It also remains superior to most (if not all) alternatives for the visual display of information. I haven’t tried Julia yet though, it also looks interesting.
• Bash/Shell scripting
  • because regardless of what people think, awk is a programming language - and because you cannot do devops without it
• SQL
  • because I needed to store things once
• Front-end programming - Astro/CSS, and a little bit of Javascript
  • you are looking at the extent of my experience with front-end programming
• perl
  • please don’t make me

• Next-generation sequencing (NGS) analyses. I have already used the following NGS-based strategies in the course of my work:
  • NGS-based variant detection (SNVs, indels, CNVs, gene fusions, neoisoforms, transposable elements…)
  • bulk RNAseq
  • scRNAseq
  • GWAS
  • ChIP-seq
  • Exosome sequencing
  • ctDNA sequencing
  • long-read sequencing
  • bisulfite sequencing
• Proteomics approaches
  • Proteomics
  • Ligandomics, in particular immunopeptidomics
  • Interactomics
• Pipeline automation and streamlining
• Reasoned infrastructure in local, HPC, or cloud-based setups
• Monitoring, compliance, documentation
• Handover
  • I truly believe this is often a heavily disregarded yet crucial part of any bioinformatics approach. Who’s going to run your tool when you’re gone ?

• Data collection and curation; feature extraction
• Model building and benchmarking, hyperparameter tuning and monitoring
• Production-ready packaging
• Deployment as local tools, distributed ones, or cloud-based, serverless ones
• Post-production monitoring

Establishing deliverables, estimating timelines, follow-up and monitoring, results and insights presentation