Thanks Andrew!

While we all share our sequencing data freely, image datasets like these are still often kept private.

I am only aware of a single other study on guppy coloration (there are many!) where images are available, and it's also from our lab (Morris et al. 2020).

I hope this can improve in the future!

Full paper now out in GENETICS:
The relationship between sexual dimorphism and intersex correlation: do models support intuition?
🔗 academic.oup.com/genetics/art...
#Evolution #QuantGenetics @GeneticsGSA

Yes I personally prefer to use “deep learning” over “AI”. Not much use for LLMs for now (but perhaps for eg meta-analyses in a couple of years?), but transformer architecture may still hold a lot of promise for sequence data.. Time will tell.

And yes, we used unets for segmentation.

Guppies! ;) I didn’t talk about it, but we also use triplet learning to do GWAS on whole patterns: doi.org/10.1038/s415.... There were quite a few posters on neural nets for image analysis and this is a domain where one can be effective with modest sample sizes.

This website seems odd, it assumes MEPs will vote along country blocks, not European or individual parties?

Thanks Will, hope you’re doing well!