Genoio has a v0.2.0 release. Performance is better across the board, beating most other libraries or matching highly performance ones (but with a single interface in Python now). Check it out mancusolab.github.io/genoio/

In #GENETICS, @epigenci.bsky.social et al. present a collection of methods and #opensource software that use the ancestral recombination graphs (ARG) to efficiently fit linear mixed model to population phenotype data, for variance component estimation and genetic prediction. 🔗 buff.ly/9svg643

Ok, bc I don't know how to relax while on vacation, I put together a Python package for reading VCF/BCF, PLINK1/2, and BGEN genotype data into NumPy or SciPy sparse matrices. It’s Rust-backed for fast reads, with block/region iteration and composable variant filters. mancusolab.github.io/genoio

This year’s Genomic History Inference Strategies Tournament includes challenges in inferring relatedness. ghist.bio. If you haven’t worked in this problem before, here’s a great resource to help you get started.

postdoc alert 🚨! NIH-funded postdoc in my lab to work on Ancestral Recombination Graphs in space. Please spread the word widely in your networks, and please reach out if you're interested! (2-3yr position, start date flexible, pay/benefits good!). details here: careers.umich.edu/job_detail/2...

Bai et al. find that structural variants explain "at least 4.7% of the common genetic variation for complex traits", meaning about 1% of trait variance (on average across many traits). www.nature.com/articles/s41...

so called molecular heritability isn't even narrow sense heritability but in fact narrower narrow sense heritability as it's at most capturing genic variance and not additive variance.

Is there a software widely used that runs whatever stochastic process (OU, BM etc etc) on a tree?

Super excited to see this out! Fantastic collaboration with Luke O'Connor and trainees Amber Shen and Xinran Wang. Thread with details will come soon, but linear ARG provide a HIGHLY efficient representation of genotype data that can be treated as a linear operator www.biorxiv.org/content/10.6...

Highly efficient genotype compression leveraging genealogical relatedness https://www.biorxiv.org/content/10.64898/2026.04.29.721594v1

Something I increasingly recognize working on LD is that correlation betweem two loci at single time section, which is an accumulation of many steps prior to it is very different from a single step correlation which is truely due to linkage.

Please share! The USC Department of Quantitative and Computational Biology is searching for a full-time, teaching-track faculty member for the teaching of QBIO undergraduate statistics and computational biology courses. usccareers.usc.edu/job/los-ange...

This technology needs a health warning, like cigarettes: overuse of AI accelerates cognitive decline. www.bbc.com/future/artic...

How to choose a "good tester" in testcrosses? 🎯🌾 ✓ Accuracy jumps most from 1 to 2 testers, though "noise" depends on dominance and pool structure. Next: Pareto framework to pick testers balancing GCA and complementarity! 🚀 Thanks to Daniel Tolhurst, @gregorgorjanc.bsky.social , Sven Weber and BASF!

Happy to share a new preprint, where we studied the expected outcomes of polygenic embryo screening in real-world IVF patients. Key conclusion: gains are very small for most infertility patients, with risk reductions under 5%. 🧵1/12 www.medrxiv.org/content/10.6...

academic.oup.com/genetics/adv... Good to see papers all presented in 2025 probgen appearing in one of the greatest venues.

Why are Dutch people so tall? This has been a long-standing debate in genetics, with earlier studies questioned due to unaccounted for confounding. Great to be involved in a new study (link below) using within-family GWAS; we confirm that height has been under positive selection.

Post processing slim outputs towards statgen applications will become so much faster 🥰🥰

A little widget to play around with the various inputs and outputs that go into soft money research lab finances on a 10-year outlook. sashagusev.github.io/PISim/

Anything interesting published over the past 1.5 months that I should catch up on?

Some slides from a recent talk on missing heritability. www.dropbox.com/scl/fi/kvogj...

Great piece of work and great to see a Drug X Genotype (a type of GxE!) being shown so cleanly. Hats off to @adamauton.bsky.social and team

How do we know that the formula is correct? I run forward simulations in SLiM. Here is my another opinionated take: mainstream benchmarks with random effects models are half-baked because they fix the genotype and only resample random effects. (6/n)

Instead, I use the term "mutational variance" because it adheres better to classic population genetic literature and is clear from the name where the variance comes from. Genetic variance is made up of many things and we should be explicit about the mechanism behind it. (3/n)

How do we obtain the correct frequency dependent formula for random effects in linear mixed models? We introduce a principled procedure on how to dump population genetics to linear mixed models. (1/n)

This is why I oppose to the convention of putting the data upfront in defining the quantities' name. It should be the mechanism behind evolution and not the data source used that should be on the name of the quantity. It should be mutational, meiotic etc variance, not SNP,WGS-variance. 7/7

journal website version ! academic.oup.com/genetics/adv...

This is truly gold.

Two post-doc post in ARG space and loads of internal and external collaboration Postdoctoral Research Associate in quantitative genomics elxw.fa.em3.oraclecloud.com/hcmUI/Candid... Postdoctoral Research Associate in quantitative genetics and breeding elxw.fa.em3.oraclecloud.com/hcmUI/Candid...

Bay Area Popgen. FREE awesome popgen conference. Registration and application for talks now live on our website!

codex w/ mutli-agents really really shines. watching it spawn several agents to take care of separate coding tasks is mind blowing.

Gave Codex an old R simulation I had lying around and it very adeptly turned it into an interactive web app with tunable parameters and visualizations. sashagusev.github.io/Stabilizing-...

flamingtempura.github.io/bibtex-tidy/... So lovely

"Monogenic no more: are all epilepsies polygenic?" by Remi Stevelink & colleagues FREE till April 1st at authors.elsevier.com/a/1mbEocQbJI...

Link to the updated version: www.biorxiv.org/content/10.1...

Our group (PI Jonathan Terhorst) is looking for a talented postdoc in computational biology, statistics, population genetics to work on quantitative genetics and ARG, as part of a larger collaboration with groups in Oxford and Edinburgh. Details to follow. (4/4)

Comparison to BOLT-LMM clearly shows the advantage of this new class of ARG-based algorithms over more traditional ones. The backbone (implicit genotype access by matmul) is the same, but ARG offers a huge improvement. (3/n)

The work is based on fast matrix access via ARG. Zhu et al. (2025) offers a similar but different algorithm in arg-needle-lib package. We found that tskit's exact algorithm is not only more accurate, but also much faster (~20-200x) than their Monte-Carlo algo. (2/n) pubmed.ncbi.nlm.nih.gov/41380687/

Our recent work "Genetic prediction with ARG-powered linear algebra" was recently accepted to Genetics and we also have a postdoc job opening related to the paper, supported by a private philanthropy for 3 years. The main additions are timing benchmarks to arg-needle-lib and BOLT-LMM. 1/n

Another Claude project: a static site that pulls in GWAS SNP data from ensemble, multiple public biobanks, open targets, gtex, eqtl catalog, and OMIM. sashagusev.github.io/gwas_lookup/

Excited for this. Teaching popgen is one of my favourite parts of my job and I've wondered about doing an online version of the class for sometime. Teaming up with @jrossibarra.bsky.social to do this is wonderful.

Following an Rcpp question 'how do I share an object / external pointer' between #Rstats and #Python, I cooked up a simple 'stopwatch' example in two repos with two packages: github.com/eddelbuettel... github.com/eddelbuettel... The Python side is on PyPi, shall I send the R side to CRAN?

Did anyone try telling an AI agent to format your ms according to the journal requirements?

A very lazy Q: Is Levy process considered commonly in phylo comparative methods? www.biorxiv.org/content/10.6...

If anyone regrets not being able to attend my recent regression modeling course you can still access the slides, exercises, and recordings with a late donation. For donation instructions see betanalpha.github.io/courses/.

#mathshy #statistics

As a (to me very enjoyable) part of this paper, we worked out what mutation-selection balance looks like in finite populations with varying degrees of inbreeding.

Preprint by Gravel et al addressing an interesting question. Given carriers of a mutation and a population-scale pedigree, can we identify the most recent common ancestor of all carriers? "20 carriers usually provide enough information to reliably identify a common ancestor 15 generations ago"

A whole Advent Calendar of Nigel Farage repeating Vladimir Putin's talking points. Brilliant. bylinetimes.com/2025/12/10/n...