iiinova — where your digital self keeps evolving

Andrew Brown @arnsbr.bsky.social · 16/06/2026

The word compute is very useful in dealing with university administration and grant applications as "I need to spend £1000 on compute" sounds professional and "I need to spend £1000 on computers" sounds like I want an Xbox

020

Andrew Brown @arnsbr.bsky.social · 08/09/2025

When I lived there I called it the rule of fun. If it was fun, alcohol, cigarettes, bacon, it cost twice as much compared to London where I was before. Not fun, rent, transport, it was cheaper. Since salaries were double, you came out on top. Good years.

040

Andrew Brown @arnsbr.bsky.social · 12/06/2025

There's an exemption for players, for the world cup and also the Olympics, but supporters are screwed

000

Andrew Brown @arnsbr.bsky.social · 26/04/2025

25 years since I was an English teacher and "bugger me it's Mrs Smith, I thought she was on holiday" is as fresh as if it was yesterday

110

Andrew Brown @arnsbr.bsky.social · 22/03/2025

If he revoked it each time in front of a mirror, watch out for those bees!

020

Andrew Brown @arnsbr.bsky.social · 28/02/2025

I think I identify as English because growing up it was the BNP that wanted to deport me. Then those sneaky racists rebranded as the EDL. But I think even in the 80s and 90s, long after Orwell, for much of England Englishness and Britishness were indistinguishable

010

Andrew Brown @arnsbr.bsky.social · 05/02/2025

Has anyone in the US thought about writing their NIH grants in Polari?

012

Andrew Brown @arnsbr.bsky.social · 14/01/2025

My interview song. Buzzin' by Asian dub foundation

010

Andrew Brown @arnsbr.bsky.social · 28/11/2024

I think it's common in economics to have these kinds of schemes in place, speaking as someone who just submitted their review a month late I guess they know the value of incentives

000

Andrew Brown @arnsbr.bsky.social · 23/11/2024

Finally get to say the words I've been dreaming of saying since I was a little lad, "today, let's go to the world's first festival devoted to pork scratchings"

000

Andrew Brown @arnsbr.bsky.social · 19/11/2024

Probably both? Fine napping methods are fairly blunt tools and assays fairly bad at capturing regulatory regions?

140

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Huge thanks again to everyone who contributed to this work: @anavinuela.bsky.social, Ewan Pearson, @theodyssee.bsky.social, Martijn van de Bunt, @hakyim.bsky.social, @moneeza-ks.bsky.social, Femke Rutters, Paul Franks, Mark Walker, Naeimeh Atabaki Pasdar, Dina Mansour Aly.

020

Andrew Brown @arnsbr.bsky.social · 13/11/2024

In summary: whole blood datasets are useful! And large reference panels can provide novel insights on the genetic of complex diseases and disease subtypes.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Here we could find genes relevant to particular processes, 𝘛𝘙𝘔𝘛2𝘈, a gene involved in beta cell apoptosis pops up with severe insulin deficient diabetes. But these subtype GWASs are underpowered: we only find this gene by using a large whole blood reference to compensate.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Ok, but BMI is a terrible metric to study T2D, and maybe disease subtypes are still something to consider. So we call our friend Dina Mansour Aly, used the summary statistic of her GWAS study in 5 T2D subtypes.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

We tested using GWAS summary statistics with BMI control, expecting fewer genes in obesity-related tissues like adipose but increased power in beta cell tissues by reducing variation. But all tissues lost genes, and there was no link with the relevance of the tissue for obesity.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

But how can we investigate specific aspects of T2D if islets alone can’t reveal beta cell dysfunction? T2D is a heterogeneous disease—some cases are driven by beta-cells dysfunction in the pancreas, others involve insulin resistance in other tissues due to obesity. Let’s check:

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

So, can we stop collecting “difficult” tissues? NO! Remember that when comparing across tissues, ~50% of the genes only showed one tissue. For example, 𝘛𝘊𝘍7𝘓2 was only found using pancreatic islets, the only tissue reporting an eQTL involving the known T2D-GWAS loci.

120

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Could we validate our genes with another approach? We conducted multiple instrument Mendelian Randomization testing to assess concordant SNP effects on both gene expression change and T2D risk, again finding more significant genes in the DIRECT study, supported by more SNPs.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

There is a lot of value on combining results from different tissues, since about half of the genes were found in only one tissue. Most of the 1,818 genes (~80%) were not reported before as related to T2D, with relevant tissues, like pancreatic islets, identifying more known genes.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

In total, across all 51 datasets, we found 1,818 genes and 7 proteins relevant for T2D, of which 404 were identified using the whole blood large dataset (DIRECT). GTEx (n<706) identified a total of 1,568, but no tissue found more than 299 genes.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Comparing relevant vs well powered reference panels, we found that sample size,not tissue relevance for the disease, drove gene discovery in T2D (A). The relationship between sample size and discovery was almost linear (B), with maybe saturation for the largest dataset (n=3,029).

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

Using the relevant tissue, such as pancreatic islets, you can pick up tissue specific effects on T2D. But pancreatic islets are hard to collect, much easier to collect thousands of blood samples, map many more genetic effects, and some will be cross tissue and relevant for T2D.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

We discuss the trade-off between sample size and tissue specificity of molecular datasets for the study of T2D, by using TWAS methods on three gene expression reference panels with sample sizes ranging from 70 to 3,029, from 51 human tissues, to find T2D causal genes.

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

I am very excited and proud to announce this preprint covering one part of the work done during my PhD! Impact of sample size and tissue relevance on Type 2 Diabetes gene identification. [Thread] www.medrxiv.org/content/10.1...

100

Andrew Brown @arnsbr.bsky.social · 13/11/2024

So my former PhD student David Davtian has produced his first paper from his PhD, and since we're so proud of it, I'm going to steal all the tweets he wrote about it and post them here: x.com/DavidDavtian...

x.com

121

Andrew Brown @arnsbr.bsky.social · 01/09/2024

Riots in UK were my inspiration

010

Andrew Brown @arnsbr.bsky.social · 31/08/2024

@vitoraguiar.bsky.social hope you didn't sign up to this app to get away from your countrymen. Because they've tracked you down! In numbers!

110

Andrew Brown @arnsbr.bsky.social · 24/08/2024

I'm quite a fan of this paper, about how the millions we've spent looking into the genetic determinants of longevity have mainly been looking at genetic determinants of pension fraud www.biorxiv.org/content/10.1...

biorxiv.org

Supercentenarian and remarkable age records exhibit patterns indicative of clerical errors and pension fraud

bioRxiv - the preprint server for biology, operated by Cold Spring Harbor Laboratory, a research and educational institution

914073