A new preprint from my team at @AdaptiveBiotech!

RT @ryanburge: Here's the share of each state that was atheist, agnostic, or nothing in particular in 2008 and 2022. In 2008, there were….

RT @LongFormMath: It’s mistakes like this that cause people to lose faith in journalism

RT @RevChrisDavis: What I was not expecting at @timkellernyc’s memorial service was that Tim chose the hymns and gave reasons why he chose….

RT @CL_Rothschild: 🚀 Exciting news! Introducing my (first!) course: "Better Data Visualizations with Svelte" w/ @newlinedotco . ~10 hours o….

RT @rolandgarros: In Novak voice: NOT TOO BAD 🤯. #RolandGarros | @carlosalcaraz

We hope that our methods will enable new types of ARG-based complex trait analyses. We think they could be particularly valuable for under-sequenced populations. Our software manual gives steps for exploring further!

There are several other pieces worth checking out in the paper:.- Benchmarking our ARG-Needle and ASMC-clust algorithms alongside other methods.- Introducing ARG-based mixed models for heritability, prediction, and association.- Higher-frequency (MAF > 0.1%) UKB association.

We validated these signals by leveraging the UK Biobank 200K whole exome sequencing release. For intuition on how an ARG inferred from array data can outperform imputation, see our Extended Data Fig. 5a as well as Kelley Harris' article.

Across 7 tested traits, our approach uncovers 134 rare and ultra-rare (MAF < 0.1%) variants after filtering for independent signals, more than twice the amount found by association of HRC-imputed variants (N = 64).

The general idea of testing genealogical structures has been around for decades. Our approach incorporates efficient linear-mixed-model association to improve power, while our ARG-Needle inference algorithm enabled this analysis to be done at biobank scale.

Kelley Harris @Kelley__Harris wrote a wonderfully accessible overview of our work which does a great job motivating genealogy-wide association:

From genotyping array data of 337K individuals, we inferred a structure called the ARG, which captures genome-wide genetic relationships. Our "genealogy-wide association" framework then treats each branch of the ARG as a possible mutation and tests these branches for association.

Our work on genealogical inference and association in the UK Biobank is out! Our ARG-Needle software package ("pip install arg-needle") is also now on PyPI: With @aabiddanda @AFGunnarsson @FergusCooper and Pier Palamara.

RT @SashaGusevPosts: Absolute tour de force from Brian Zhang and Palamara Lab: efficiently reconstructing ARGs to estimate heritability, in….

RT @random_walker: It feels like we're at a critical moment for AI and civil society. There's a real possibility that the last 5+ years of….

RT @W_R_Chase: I made a course 🎉🎉🎉 I stuffed everything I know about #dataviz design into this course, all with practical examples in #rsta….

Cool chart from

RT @philippschmitt: New research-y project: Blueprints for Intelligence, a visual history of artificial neural networks from 1943 to 2020.h….