iiinova — where your digital self keeps evolving

Reposted by Elizabeth Atkinson

From the Labs at Baylor College of Medicine @bcmfromthelabs.bsky.social · 22/07/2026

Dr. E. Atkinson @egatkinson.bsky.social et al have developed a new computational method that enables scientists to more accurately identify #geneticChanges linked to #disease. @natgenet.nature.com @bcmhouston.bsky.social #TexasChildrens @bcmgenetics.bsky.social www.bcm.edu/news/new-app...

033

Elizabeth Atkinson @egatkinson.bsky.social · 20/07/2026

The big picture: admixed and related samples are common in biobanks and cohorts around the world, but researchers have lacked a dedicated tool to analyze both features together. Tractor-Mix opens the door to more powerful GWAS without excluding relatives or losing ancestry-specific insight.

011

Elizabeth Atkinson @egatkinson.bsky.social · 20/07/2026

In applications to UK Biobank, Yale-Penn, and the Mexico City Prospective Study (where 71% of participants had a close relative in the study) Tractor-Mix maintained calibration, recovered established associations, and found ancestry-specific signals that standard approaches could obscure.

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/07/2026

In simulations, Tractor-Mix controlled false positives and generated reliable ancestry-specific effects. It gained power when effects differed across backgrounds, particularly for signals on lower-proportion ancestry; as expected for a 2-d.f. test, it lost power when marginal effects were identical.

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/07/2026

Tractor-Mix uses ancestry-specific genotype dosages plus a genetic relationship matrix. A joint 2-d.f. test supports locus discovery, while ancestry-specific effect sizes and P values show which ancestry background drives an association - useful for fine-mapping and risk prediction.

110

Elizabeth Atkinson @egatkinson.bsky.social · 20/07/2026

Why was this needed? GWAS methods have been designed to account for admixture or familial relatedness, but not both together. In many cohorts and in biobanks, these sources of structure often coexist, creating a need for a unified model that handles both without excluding relatives.

100

Elizabeth Atkinson @egatkinson.bsky.social · 16/03/2026

@nirav-shah.bsky.social's Tractor Workflow paper is out early access in Bioinformatics today! Check out his thread quoted here for a full bluetorial on the contents, and see the link below for the final version: academic.oup.com/bioinformati...

academic.oup.com

Tractor Workflow: A Scalable Nextflow Framework for Local Ancestry-Aware Genome-Wide Association Studies

AbstractMotivation. The routine exclusion of admixed individuals from traditional Genome-Wide Association Studies (GWAS) due to concerns about spurious ass

010

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

Thanks to all of our SMaHT colleagues and especially to @sedlazeck.bsky.social who led the hackathon which spawned the prototype of this pipeline!

000

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

MosaicSim offers a realistic, scalable approach for assessing detection limits, with immediate applications to large sequencing efforts including those within the SMaHT Network, which was the springboard for this work.

100

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

A key (surprising) result was that ultra-high coverage (300×–450×) yields diminishing returns for mosaic variant detection. In many settings, 150× coverage performs comparably or better, highlighting opportunities for cost-effective study design.

120

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

Using MosaicSim, we benchmarked DRAGEN and found strong VAF- and depth-dependent performance limits. Sensitivity decreases sharply at low VAF, especially in complex genomic regions.

100

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

Detecting mosaic variants is challenging due to low VAFs and real sequencing noise. MosaicSim layers user-defined variants directly onto empirical WGS data, preserving true read-level properties while providing a controlled ground-truth set for benchmarking.

100

Elizabeth Atkinson @egatkinson.bsky.social · 09/12/2025

We are pleased to share our new preprint introducing MosaicSim, a framework for generating realistic mosaic variants! Mosaic variants - mutations present in only a subset of cells - are crucial for development, disease, and cancer, but are notoriously hard to call. www.biorxiv.org/content/10.6...

biorxiv.org

MosaicSim: A Novel Mosaic Variant Simulator Reveals Diminishing Returns of Ultra-High Coverage for Mosaic Variant Detection

Genetic mutations within select cells of a tissue, termed mosaic variants (MV), are being increasingly recognized for their role in human disease. This growing interest underscores the need for specia...

140

Elizabeth Atkinson @egatkinson.bsky.social · 12/10/2025

A fun lab outing to the zoo ahead of conference season! 🦒

020

Elizabeth Atkinson @egatkinson.bsky.social · 10/10/2025

So since we only include >0.1% MAF variants in this article we can't address ultrarare, but check out Supp Fig 3; when comparing ancestry-specific AFs many variants deviate from the 1:1 line. We plotted this on the log₁₀(AF) scale to help magnify the low-frequency range.

000

Elizabeth Atkinson @egatkinson.bsky.social · 10/10/2025

To limit the noise from ultra-rare alleles we only looked at variants ≥0.1% MAF. Totally appreciate that's still quite low frequency, but even with that filter, we still saw the noted ancestry-specific frequency differences.

000

Elizabeth Atkinson @egatkinson.bsky.social · 10/10/2025

Great point; we thought about that too! Pragati stratified by whether variants were monomorphic or not to capture at least that aspect, but you’re right that the impact depends on where a variant sits on the SFS. Rare ones can show big fold-changes but small absolute shifts.

000

Reposted by Elizabeth Atkinson

From the Labs at Baylor College of Medicine @bcmfromthelabs.bsky.social · 06/10/2025

Texas Children's/Baylor College of Medicine Researchers Create Groundbreaking Tool to Improve Accuracy of #GeneticTesting @egatkinson.bsky.social @bcmgenetics.bsky.social @bcmhouston.bsky.social #TCHResearchNews #TexasChildrens @natcomms.nature.com tinyurl.com/jj6kyrrv

051

Elizabeth Atkinson @egatkinson.bsky.social · 06/10/2025

Thrilled to share our new @natcomms.nature.com paper on local ancestry informed allele frequencies in gnomAD, which are live now on the browser! Check out my stellar PhD student @pragskore.bsky.social’s Bluetorial on how this brings finer detail to variant interpretation 🧬🖥️

1144

Elizabeth Atkinson @egatkinson.bsky.social · 13/09/2025

Delighted to amplify my talented PhD student’s work! Check it out for a great way to streamline and harmonize Tractor analyses.

062

Elizabeth Atkinson @egatkinson.bsky.social · 23/07/2025

Thanks for the interest! The tutorial code is available to download as supplemental information of the paper, and has been deposited as a community workspace in the All of Us Researcher Workbench.

020

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

In summary, we present a replicable training model that empowers early-career researchers - including and especially those new to computational genomics - to responsibly leverage large-scale biobank data into their research programs and teaching.

100

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

From years 1–3, training outcomes reported by scholars to stem directly from this training included: 📊 17 conference presentations 🔬 Multiple funded research grants 🎓 Numerous genomics modules added in undergrad courses 🤝 Sustained collaborations across institutions

100

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

During the summit, scholars used real short-read WGS data to: • Prepare phenotypes & covariates • Run GWAS via Hail • Visualize results with PCA, Manhattan & QQ plots • Manage compute costs All in ~4 hours with no prior coding required.

100

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

Our training was part of the All of Us Biomedical Researcher Scholars Program through @bcmgenetics.bsky.social focused on mentoring early-stage faculty in genomic data science. The curriculum launches with an intensive Faculty Summit, where scholars get hands-on experience working with genomic data.

100

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

Access to big genomic data is growing, but parallel access to skills needed to use it hasn’t kept up. We created an accessible, cloud-based genomic analysis training bootcamp using real All of Us data, Jupyter notebooks, and the Hail framework to lower the barrier for early-career researchers.

100

Elizabeth Atkinson @egatkinson.bsky.social · 22/07/2025

🚨 New perspective piece in @ajhgnews.bsky.social! 🚨 We developed a hands-on training resource for large-scale genomic data analysis in the All of Us Researcher Workbench, now published here:

1138

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

Tractor-Mix builds on Tractor’s strengths to detect ancestry-enriched signals while adding power and robust false-positive control for relatedness via a GRM. By modeling both admixture and relatedness, it overcomes key GWAS barriers and enables more accurate, representative genomic discovery.

020

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

Tractor-Mix uses ancestry-specific genotypes as predictors, outputting ancestry-specific effect sizes and P values. We benchmark our new tool in simulations and apply it to multiple admixed cohorts (including UKBiobank and Mexico City Prospective Study), uncovering signals missed by standard GWAS.

120

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

In this work, we introduce Tractor-Mix, a new GWAS method that extends Tractor to handle related admixed samples. It combines a mixed model framework (like GMMAT) with local ancestry-aware genotypes (like Tractor) in a 2 d.o.f. test.

120

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

As biobanks and global cohorts grow, so does the inclusion of admixed individuals with close or cryptic relatedness. This introduces the statistical challenge of two interwoven sources of stratification: admixture and relatedness, which are rarely handled together.

120

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

We previously developed Tractor, a local ancestry-aware GWAS method that’s been widely used to uncover ancestry-enriched signals and refine genetic architecture in admixed populations. But Tractor (being a GLM) only works on unrelated samples, limiting its use in many real-world datasets.

120

Elizabeth Atkinson @egatkinson.bsky.social · 09/06/2025

We're excited to introduce Tractor-Mix, our new method for GWAS in admixed cohorts with relatedness, led by the fantastic @doubletaotan.bsky.social! Read the full preprint here: www.medrxiv.org/content/10.1... Thanks to all our amazing collaborators who helped make this work possible!

1123

Elizabeth Atkinson @egatkinson.bsky.social · 09/04/2025

Check out my stellar PhD student, Pragati's talk on our work generating local ancestry informed frequency estimates in gnomAD as part of the prestigious Emerging Genomic Scientist Symposium next week! Congrats on being selected for this amazing event!

medschool.ucla.edu

Human Genetics | Genomic Scientist Fellows | UCLA Medical School

The Emerging Genomic Scientist Fellows Program is a cornerstone of justice, equity, diversity, and inclusion initiatives in the Department of Human Genetics.

051

Elizabeth Atkinson @egatkinson.bsky.social · 03/04/2025

I'm delighted to be part of this symposium, put on by University of Pennsylvania Perelman School of Medicine, and led by @bpasaniuc.bsky.social and @sarahtishkoff.bsky.social. See you in a few weeks! upenn.co1.qualtrics.com/jfe/form/SV_...

084

Elizabeth Atkinson @egatkinson.bsky.social · 02/04/2025

👏 Huge thanks to all our amazing LAGC collaborators! Special shoutout to Estela Bruxel and Diego Rovaris for leading this crucial work, and of course @janitzamontalvo.bsky.social and @giustilab.bsky.social for co-founding the LAGC and co-leading alongside myself. 💪

021

Elizabeth Atkinson @egatkinson.bsky.social · 02/04/2025

🔍 Why does this matter? Most psychiatric GWAS are still Euro-centric, limiting the relevance of genetic findings across populations and ancestries. Latin America’s rich genetic, environmental, and cultural diversity presents a unique opportunity to refine genetic discovery & improve global research

111

Elizabeth Atkinson @egatkinson.bsky.social · 02/04/2025

In this review, we: 🌎 Examine the current state of psychiatric genetic studies in the region 💰 Highlight key challenges in data generation, analysis, & funding 🧬 Showcase emerging opportunities for more representative genomic research 🤝 Call for greater collaboration & investment

111

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

Check out our preprint for more details! Huge thanks to All of Us participants & our fantastic team! Let’s continue improving polygenic prediction for all 🔬🧬

doi.org

000

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

7/ Looking ahead: ➡️ As sequencing costs drop, WGS could become standard for PGS. ➡️ More representative reference panels & ancestry-aware methods are crucial for equitable genetic prediction. ➡️ Future work should continue to explore improved LD modeling across populations.

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

6/ Implications for Genetic Research: Choosing between arrays & WGS for PGS depends on: ✔️ Trait polygenicity ✔️ Target population ancestry ✔️ PRS method used ✔️ Budget & computational resources

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

5/ Computational and Cost Considerations: ➡️ Arrays are faster & cheaper (~$40-$135/sample). ➡️ WGS is expensive (~$600/sample) but captures more variants. ➡️ PRS-CS improves WGS performance but is computationally intensive (weeks vs minutes for C+T).

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

4/ Key Findings – PRS-CS (LD-Informed Method): ➡️ WGS-based PGS outperformed arrays for most traits, particularly highly polygenic ones (e.g. height). ➡️ The relative advantage of WGS was smallest in AFR populations, emphasizing persistent disparities in genetic prediction across populations.

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

3/ Key Findings – Clumping & Thresholding (C+T): ➡️ WGS did NOT consistently outperform arrays. ➡️ This was likely due to the severe variant reduction from LD clumping (~95% of WGS variants removed). ➡️ Lower allele frequencies in clumped WGS variants relative to arrays also played a role.

110

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

2/ Study Design: We compared PGS derived from arrays vs WGS across 2 methods, 10 complex traits, and 3 major genetic ancestry groups (European [EUR], African [AFR], Admixed American [AMR]) in the same people in the All of Us Research Program. PGS were trained using Pan-UK Biobank summary statistics.

100

Elizabeth Atkinson @egatkinson.bsky.social · 20/03/2025

1/ Background: PGS are widely used to estimate genetic predisposition to complex traits. However, the choice of genotype discovery approach (arrays vs. WGS) could have important implications for predictive performance, portability, and computational efficiency.

110