Thank you!

Enormous thanks to our supervisors @stephansanders.bsky.social and Tomasz Nowakowski, our co-authors, the @simonsfoundation.org for supporting this work and the Autism Brain Net. Most importantly, thank you to the donors and their families, without whom this work wouldn't have been possible.

Overall:
• strongest signal in genetically diagnosed donors
• weaker but moderate concordant signal without diagnosis
• convergence on neuronal regulatory programs
• RFX3 as a hub linking chromatin → transcription → activity-dependent genes

Finally, these language-associated regulatory changes are enriched in recently evolved genomic regions linked to human language.

We also looked at verbal vs non-verbal status (small Ns, interpret cautiously):
• effects are again neuron-specific
• RFX3-associated regulatory activity varies within autism by language phenotype.

This RFX3 → IEG axis is strongest in the same neurons driving the case-control signal for expression and motif acccessibility (L4/5 IT).

So we asked: what’s going on with IEGs in our data?

IEGs are enriched among DEGs (VIP + L4/5 IT), but many are downregulated.

Despite increased RFX3 accessibility/expression, activity-dependent programs may be altered (possibly compensatory).

RFX3 is also an autism gene implicated through rare variant burden (www.nature.com/articles/s41...) and a regulator of activity-dependent transcription of immediate early genes (IEGs) (www.biorxiv.org/content/10.1...).

Another clear signal: a recurring regulatory signal points to RFX3 (my new favourite gene).

We see:
• increased motif accessibility (DAM; especially in excitatory neurons, L4/5 IT)
• increased expression in NDD/autistic donors
• consistent effects across comparisons

So:
• not necessarily different biology 🧠
• but different effect sizes 📊
→ consistent with a spectrum from high-penetrance variants to more distributed risk

When we look genome-wide (not just significant DEGs), the signal is moderately correlated (especially given how heterogeneous diagnostic variants are), with the strongest concordance in inhibitory neurons. We were careful to use different controls to avoid an inflated correlation.

First question then: is the signal in donors without a genetic diagnosis the same but smaller than in donors with a genetic diagnosis? Or are they fundamentally different? 🤔💭

One of the clearest signals: gene expression differences between cases and controls are much stronger in donors with a known genetic diagnosis than those without. The signal is strongly neuronal, especially deep-layer excitatory neurons (L4/5 IT).

📣 New preprint 🧵: we profiled post-mortem human BA22 (speech cortex) in autism across 100 donors using single-nucleus multiomics (~500,000 nuclei), integrating RNA + chromatin accessibility. Shout-out to my co-first authors Varun Suresh and Yuhan Hao: www.biorxiv.org/content/10.6...

Super excited that our two companion papers on saturation genome editing (SGE) of RNU4-2 and discovery of a novel recessive neurodevelopmental disorder (NDD) were published yesterday 🥳

SGE experiment: www.nature.com/articles/s41...
Recessive NDD characterisation: www.nature.com/articles/s41...

🧵

Targeted BDNF upregulation via upstream open reading frame disruption To understand the relative contributions of 5ʹ UTR elements to translation output, we performed a comprehensive analysis of upstream open reading fram…

Latest research from my group. We activated BDNF protein expression by disrupting a repressive uORF using a CRISPR base editing strategy: www.sciencedirect.com/science/arti...

#OxfordPaediatrics #IDRMOxford #BDNF #uORF #CRISPR #BaseEditing

Why do some individuals defy their polygenic score?

In the largest study of its kind (402k UKB individuals; 7 continuous traits + 3 diseases), we asked: If your phenotype deviates from common-variant polygenic score prediction, what's driving that difference?

www.medrxiv.org/content/10.6...

Congratulations, Hilary!! Well deserved!

Genotype-level quality control substantially reduces error rates in population-scale whole-genome sequencing Population-scale whole-genome sequencing data will contain many individual-level genotype errors, even after allele-level quality control (QC). We establish the need for genotype-level QC using UK Bio...

New paper on everyone’s favourite topic, QC!
We show why you should do genotype-level QC on your WGS data

www.biorxiv.org/content/10.1...

Very real quotes about this paper -
“The most exciting, mind-blowing paper of the year!”
“On a par with Fisher 1918”
“I read it every night. Just so beautiful”

🚀 Very excited to share the first major work from my PhD!!

We combined MPRA and CRISPRa in excitatory neurons to test and validate cis-regulation therapies for hundreds of haploinsufficient neurodevelopmental disorder genes. 🧬🔬

www.biorxiv.org/content/10.1...

First direct evidence links gene to fat storage, hinting at new obesity therapies A new study, involving researchers at the Big Data Institute, has shown for the first time that a little-known gene, SLTM, plays a direct role in how fat is stored inside human cells. While large popu...

Read about our recent paper here: www.bdi.ox.ac.uk/news/first-d...

Combining evidence from human genetic and functional screens to identify pathways altering obesity and fat distribution Overall and tissue-specific fat accumulation are associated with altered risk of cardiometabolic disease and mortality. By combining exome-wide association analysis of traits related to obesity and fa...

✨ NEW PUBLICATION ✨

We combined large-scale human genetics with CRISPR-Cas9 editing in fat cells to identify genes linked to fat accumulation.

Check out the full study, now published in AJHG! www.cell.com/ajhg/fulltex...

New app deadline August 1st 📣

If you're interested in human genetics, stats, big data, and/or ML, this is in an incredible opportunity 🧬💻

✨Fully funded ✨ PhD at Oxford with @astheeggeggs.bsky.social, available to international students 🌎🌍🌏, embedded in the new and exciting SMARTBiomed Centre 🤩

Incredible opportunity for international and domestic students to do a PhD in stat gen at Oxford in a dynamic, engaging, fun and supportive environment! 🤩🧬🎓

Apply by June 30th for an October 2025 start!

A flyer advertising four talks and one poster: Yuyang Chen - 11:30am Saturday 24th; talk (C01) - De novo variants in small open reading frames harbour new rare disease diagnoses Anthony McGuigan - 6:45pm Saturday 24th; talk (C09) - Gene knockouts across 120,404 individuals for novel rare disease gene discovery François Lecoquierre - 7:15pm Saturday 24th; talk (C09) - A map of predicted pseudoexons in human genes Hyung Chul Kim - 1pm Sunday 25th; poster (P18.006.A) - Rare variant association study reveals small open reading frames (smORFs) as novel regulators of cardiometabolic diseases Nicky Whiffin - 10:30am Monday 26th; talk (C29) - Saturation genome editing of RNU4-2 reveals distinct dominant and recessive neurodevelopmental disorders

It's #eshg2025 #eshg25 time 🥳

Here is where you can catch the team over the next few days.

Please go and say hi!

Genome-wide association meta-analysis of age at onset of walking in over 70,000 infants of European ancestry - Nature Human Behaviour A genome-wide association study of age at onset of walking in over 70,000 infants found 11 significant loci. Age at onset of walking showed SNP heritability of 24%, a reliable polygenic score and gene...

Our gene discovery project on age at onset of walking is out today! Find the paper here www.nature.com/articles/s41... Huge thanks to the cohorts, collaborators & coauthors, in particular first author Dr Anna Gui 1/n

Very excited about this important piece of work with @angelicaronald.bsky.social characterising gene associations with age of walking onset. We also show using dHCP data that these are related to volume in key brain motor regions (like the deep grey nuclei and cerebellum) even as a newborn infant

We have two vacancies for Junior Research Fellows to join the CPM team next academic year. Please repost and send to anyone who might be interested. shorturl.at/DOV5s

Delighted to share that the revised version of this work on the genetic basis of infertility is now out in @naturegenet.bsky.social ! Give it a read and let us know what you think - rdcu.be/ehEgG 🧬💻

This is going to be a great lecture, do come along to the Sheldonian this afternoon to hear @philipcball.bsky.social