Hrushikesh Loya
@hrushikesh_loya
Statistics PhD at @UniOfOxford advised by @simon_r_myers and @pierpalamara
Excited to share our paper, on efficient variance-component estimation in ARGs, following-up on method developement by @AliPazokit and @sr_sankararaman and my PhD work. We introduce ARG-RHE to estimate narrow-sense h2 and perform region-based association testing using an ARG.
Fast variance component analysis using large-scale ancestral recombination graphs biorxiv.org/cgi/content/sh… #biorxiv_genetic
Two new chapters from my open-access textbook in human population genetics are now online: - Population structure: I. Ancestry estimation - Population structure: II. More about admixture web.stanford.edu/group/pritchar…
I'm delighted to release the first half of my new open-access online textbook in human population genetics: web.stanford.edu/group/pritchar…
Exciting work from @leo_speidel !!
We’re excited to share Twigstats, boosting f-statistics power for fine-scale ancestry modelling by using genealogies, compatible with admixtools2. Application to early medieval Europe reveals major mobility from and into Northern Europe and Scandinavia! biorxiv.org/content/10.110…
Human ancestors may have survived a brush with extinction 900,000 years ago | Science science.org/content/articl…
great work, @fuopen
New manuscript on biorxiv, led by the wonderful Sile Hu! We show underlying phenotypic effects are highly conserved across people from different ethnic groups living in the UK, despite the apparent non-portability of polygenic predictions. biorxiv.org/content/10.110…
Critics of polygenic risk scores need to work on new material ;) evaluating PRS together with established risk factors and clinical models is now standard procedure. A short 🧵:
#Statistics challenge of the day to developers of polygenic risk scores/biomarkers: 1st show that the new marker cannot successfully be predicted from already known data (risk factors, signs, symptoms, age, dz severity, standard blood panel, comorbidities).#TranslationalResearch
🧮 I finally spent some time learning what exactly Neural Tangent Kernel (NTK) is and went through some mathematical proof. Hopefully after reading this, you will not feel all the math behind NTK is that scaring, but rather, quite intuitive. lilianweng.github.io/posts/2022-09-…
Great ideas from @emidup et al
By treating functions parameterized by neural nets as data points, called functa, we introduce a framework for tackling several deep learning tasks, including diffusion models on NeRF and 3D shape inference! 💫
Introducing Flamingo 🦩: a generalist visual language model that can rapidly adapt its behaviour given just a handful of examples. Out of the box, it's also capable of rich visual dialog. Read more: dpmd.ai/dm-flamingo 1/
Great work from @notmilad and team!! 😍
Happy to share our latest work, where we use meta-gradients to prune networks *before training*. Check out the paper and code to see how we did it and how this connects to the lottery ticket hypothesis!✂️🍀#ICLR2022 📄 arxiv.org/abs/2202.08132 👨💻 github.com/mil-ad/prospr 🧵1/6👇
Great talk by @sebnowozin on his emotional rollercoaster of being a Bayesian. I think the reverend would approve. youtu.be/xRqjWoQNd4Q
Great article in @CNN about our Science paper, just published today: amp.cnn.com/cnn/2022/02/24…
Another story that reads like sci-fi. Before, we saw player of games, advances in quantum chemistry, weather forecast, protein folding, mathematics. This time, DeepMind had a breakthrough in using AI to control nuclear fusion!
Today in @nature, with @EPFL, the first deep reinforcement learning system that can keep nuclear fusion plasma stable inside its tokamaks, opening new avenues to advance nuclear fusion research. Paper: dpmd.ai/fusion-paper