Thread by @ewanbirney

Done a genetics, statistics, CS, image analysis or physics PhD? Want to do a postdoc? Love biology and genetics? Want to understand the limit of genetics, using genetics? Want to work internationally in UK, Germany and Japan? This postdoc is for you: embl.de/jobs/searchjob…

You will be based @emblebi which is part of the international treaty organisation @embl. This is a European science organisation; you will live in the UK but can come from any country worldwide. We work in an @ERC_Research funded synergy grant with @WittbrodtLab on Medaka fish

Medaka fish are japanese rice paddy fish, and we work closely with @Naruse_kiyoshi from NIBB in Nagoya Japan. Medaka are unique in vertebrates in that there is a protocol to inbreed individuals from the wild which is successful in around 1 in 2 attempts.

Using this we've set up the first vertebrate wild-derived inbred panel (the MIKK panel). Mapping this to human genetics, it is as if we are able to replicate a *population* of individuals in a studies again and again. This is common place in Arabidopsis and Drosophila

(In mammalian terms this is a bit like RILs - eg, BxD lines or Collaborative Cross in mice; but unlike these populations the variation in the MIKK panel really looks like it is drawn from the wild Medaka population - ie, looks more like human genetics)

We have done the genome sequencing and have initial intensive phenotyping happening - it is clear this will be a very powerful system. As a postdoc you will lead the analysis of these, and be co-first author on resulting papers.

I get a real tingle about being able to *explore* and *test* key aspects of phenotype in this system, such as the impact structured environment, GxE and just understanding variance in phenotypes.

These are key concepts in human studies but we nearly always have to infer this via statistical models in observational studies ("cohort studies") - we cannot by definition easily challenge or test or explore these models as the models do the inference.

We tend to choose the simplest models which are adequate (which is fine) and even in nested models, good modelling often does not want to "spend" information/model on the rare events. However, in the "tails" of these simpler models is precisely where we want to be accurate

This goes from the routine - do you fit a genotype model or an allelic model - to sophisticated - What about compound het / allelic hetreogenity - local GxG, or are GxE effects likely to be present on allele basis or a specific genotype?

In human studies one can argue, speculate and then ultimately... use simple, robust models. We can never really change or challenge this approach. But - in Medaka fish, which is just as good a vertebrate as humans - we can specifically explore all these things.

So - how much variance in a phenotype is due to structured environment (eg, temperature, calorie intact) vs "impossible to measure" environmental variation? Are there GxE effects? Do you need an allelic or a genotypic model to fit them well?

How does allelic heterogeneity work in practice with multiple haplotypes and alleles? How does that interact with GxE effects? and the phenotypes here are analogous - perhaps even "orthologous" - to human phenotypes - Heart beat. Height/Length. Fear response.

I am *super* excited about this project which continues to build and build. Apply and join us! embl.de/jobs/searchjob…