Popular in Life Sciences

Biology

Plasticity and microevolution of diatoms

University of Technology Sydney Martina Doblin

Life Sciences

The Mouse Cortical Connectome Maintains Specificity by Distinct Connectivity Profiles

Monash University Răzvan Gămănuţ

Life Sciences

Neural control of three-dimensional gaze shifts

York University

Biology

Quantifying nectar production by flowering plants in urban and rural landscapes

University of Bristol

Biology

Morbillivirus cross-species infection: is there a risk for humans?

Queens University Belfast

Life Sciences

Direct Fit to Nature

Princeton University

Biology

The Antidepressant Effects of Lactate

King Abdullah University of Science and Technology

Analysis of rare coding variants reveals novel genetic risk factors for type 2 diabetes

View more insights in Sciences - Life Sciences - Biology - Natural Sciences

View more figures from University College London

View the page of David Curtis

Read the Study

David Curtis explores findings from a previous study, utilizing an expanded sample of 470,000 exome-sequenced individuals. The aim was to identify genes linked to type 2 diabetes (T2D) and examine the distribution of rare variants within these genes.

Image courtesy of interviewee. November 30, 2023

Generating Workspace

Export as PDF

This video video discusses research involving the UK Biobank, which comprises half a million volunteer participants in the UK who have provided extensive health data, tissue samples, and undergone various assessments, including DNA sampling. This video video study focuses on exome sequencing, allowing researchers to read every DNA base in about 20,000 genes, a method that extends beyond traditional genome-wide association studies, which often examine common genetic variants predominantly outside of coding regions.

The author explains the significance of identifying rare genetic variants that could lead to diseases such as type 2 diabetes. The analysis involved comparing exome sequences of individuals with type 2 diabetes against those without it to find damaging genetic variants. Out of 20,000 genes examined, three genes were highlighted—two were previously known to affect diabetes risk (GCK and HNF4A), while one new gene (GIGYF1) was identified as a potential risk factor.

However, the author emphasizes that very few individuals with type 2 diabetes carry these rare variants, suggesting that less than 1% possess a genetic mutation that directly causes their condition. As such, the expectation that personalized medicine can reliably identify specific genetic causes for diseases and inform treatment is challenged. The author concludes that while these genetic studies may help advance the understanding of disease mechanisms, they do not yet translate to effective personalized treatment approaches for the vast majority of patients with type 2 diabetes.

Language:

Export as PDF

Export as PDF
Here are some articles that are similar to the video you , focusing on similar themes of genetic research, the UK Biobank, exome sequencing, and the implications for personalized medicine in relation to conditions like type 2 diabetes:

1. **"Understanding the Genetic Basis of Type 2 Diabetes: Insights from the UK Biobank"**
- This article discusses findings from the UK Biobank regarding genetic factors linked to type 2 diabetes and how entire genome sequencing provides insights into both common and rare variants contributing to the disease.

2. **"Exome Sequencing and Its Role in Uncovering Rare Genetic Variants Associated with Metabolic Disorders"**
- The focus of this article is on how exome sequencing is used to identify rare genetic variants and how these findings influence our understanding of metabolic disorders, including diabetes.

3. **"The Impact of Rare Genetic Variants on Type 2 Diabetes: Lessons from Exome Studies"**
- This article delves into rare genetic variants discovered through exome sequencing and their implications for type 2 diabetes risk, along with challenges in translating genetic data into clinical practice.

4. **"Personalized Medicine in Diabetes: Are We There Yet?"**
- The author examines the current state of personalized medicine in diabetes treatment, emphasizing the difficulties of linking genetic findings to practical treatment options.

5. **"The Role of the UK Biobank in Genetic Research: A Treasure Trove of Data"**
- An overview of the UK Biobank's contributions to our understanding of genetics and health, including discussions on the methodologies used such as DNA sampling, exome sequencing, and the importance of studying both common and rare variants.

6. **"Rare Variants in Diabetes: Understanding Their Role Through Exome and Genome Sequencing"**
- This article discusses research aimed at understanding how rare genetic variations contribute to the risk of type 2 diabetes, showcasing new findings and ongoing challenges in personalized treatment.

7. **"Exome Sequencing in Clinical Research: Implications for Public Health"**
- Explores the broader implications of exome sequencing in public health, particularly regarding chronic diseases like diabetes, and discusses the potential pitfalls of overinterpreting genetic data.

These articles can help deepen your understanding of the research landscape related to genetic variations in diseases like type 2 diabetes and the role of large-scale biobanks like the UK Biobank in advancing this research.

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.000 --> 00:02.070
There's this sample called the UK Biobank,

00:02.070 --> 00:04.565
which is half a million volunteers,

00:04.665 --> 00:07.805
in Britain who've given their health information

00:07.945 --> 00:10.965
and tissue samples and undergone various assessments.

00:11.425 --> 00:13.485
And one of the things they've done was to give a, a sample

00:13.665 --> 00:14.725
for the DNA

00:15.425 --> 00:19.925
and that had been the subjects of many,

00:19.925 --> 00:23.085
many studies called genome-wide association studies using,

00:23.625 --> 00:26.885
a million markers spread across all 23 chromosomes.

00:27.265 --> 00:29.245
Um, of, and these are markers

00:29.245 --> 00:32.125
where there's common variation in people's genomes.

00:32.355 --> 00:33.765
Sometimes they have 1D

00:33.765 --> 00:34.925
NA based, sometimes they have another.

00:35.185 --> 00:38.645
And there've been lots and lots of these studies, done

00:38.865 --> 00:40.605
before these genetic studies.

00:41.155 --> 00:44.445
What was new was that rather than just,

00:44.445 --> 00:46.525
have these common markers,

00:47.225 --> 00:50.845
the UK bank bank had organized for these, this DNA

00:50.845 --> 00:52.445
to be what's called exome sequence,

00:52.445 --> 00:55.165
which means you read every single DNA base

00:55.545 --> 00:56.765
in every single gene.

00:56.805 --> 00:58.365
So you've got 20,000 genes.

00:58.705 --> 01:01.005
You read every single base in every one of those genes.

01:01.465 --> 01:04.365
And that really has, that differs from the previous Genome

01:04.365 --> 01:06.605
Wide association studies in, in two ways really.

01:06.985 --> 01:10.045
One is you are looking at variants that are actually

01:10.145 --> 01:12.965
inside genes that are the kind of important bits

01:12.965 --> 01:14.485
of the genome, the one

01:14.485 --> 01:16.725
or 2% which is actually coding for proteins

01:16.725 --> 01:17.965
and kind of really doing all the work.

01:18.265 --> 01:19.845
So you're looking in the important places,

01:20.235 --> 01:22.085
whereas genome-wide association studies,

01:22.085 --> 01:24.365
their markers are spread across all the chromosomes.

01:24.405 --> 01:27.165
A lot of 'em are not in genes at all, nowhere near genes.

01:27.425 --> 01:29.685
We dunno what they do. So you're looking at the genes.

01:29.685 --> 01:31.525
And the other thing is, is that

01:31.525 --> 01:34.405
because you are reading every single base in

01:34.825 --> 01:36.205
how many people, thousands

01:36.205 --> 01:38.485
and thousands of people, instead of just looking

01:38.625 --> 01:41.925
for common variation, you can see if somebody has a very,

01:41.925 --> 01:44.965
very rare variance that maybe only they have

01:44.985 --> 01:47.045
or mainly only half a dozen people have.

01:47.305 --> 01:49.925
So you find rare variance in genes.

01:50.465 --> 01:53.445
And the difference between rare variants

01:53.445 --> 01:56.485
and common variants is common variants.

01:56.705 --> 01:58.805
We all have lots of variation in code

01:58.805 --> 02:01.165
and it has almost no effect on, on us.

02:01.185 --> 02:02.365
It has tiny effects.

02:03.115 --> 02:05.215
And if you think about it, it's almost sort

02:05.215 --> 02:06.455
of common sense really

02:06.455 --> 02:07.935
with hindsight, that that would be the case.

02:08.245 --> 02:11.175
Because if anything had a very, if there's a change

02:11.175 --> 02:13.175
that had a very big effect, made us much taller,

02:13.285 --> 02:16.255
made us much smarter, whatever it was then,

02:16.835 --> 02:20.255
or made us much more sick, any big effect,

02:20.835 --> 02:22.335
you get evolution happening

02:22.435 --> 02:26.255
and it gets either, if it's advantageous, everybody gets it.

02:26.275 --> 02:28.295
And if it's disadvantageous, then nobody gets it.

02:28.715 --> 02:31.135
So the common variation that all

02:31.135 --> 02:32.975
around doesn't really do very much,

02:33.475 --> 02:37.535
but occasionally you can have a, a rare variant will occur

02:37.835 --> 02:39.855
and it will survive in the population for a while,

02:39.855 --> 02:41.855
even if it's quite harmful, even if it kind

02:41.855 --> 02:43.575
of really increases your risk

02:43.575 --> 02:45.015
of getting a disease such as diabetes.

02:45.835 --> 02:49.465
And so what had happened was you get back, biobank as said,

02:49.465 --> 02:53.225
organized for, it was at that time 200,000 exome sequences

02:53.285 --> 02:55.785
to be released from their volunteers.

02:56.485 --> 02:58.545
And this gave the opportunity to look at these

02:58.545 --> 02:59.765
200,000 exomes

03:00.105 --> 03:03.725
and see if there are any very rare variants which might have

03:03.995 --> 03:07.365
very big effects in terms of increasing risk of, of diabetes

03:07.545 --> 03:09.285
or one of a number of other diseases.

03:09.745 --> 03:11.685
So this was what I was going into to find,

03:11.685 --> 03:14.165
and the i the point being that if you can then find

03:14.555 --> 03:16.125
that if there's a rare variant,

03:16.125 --> 03:18.685
which has a big effect in terms of damaging a gene

03:19.065 --> 03:20.885
and that gives you a disease like diabetes,

03:21.475 --> 03:23.205
then you can say, well, that's, you know,

03:23.205 --> 03:25.165
potentially you'll say, that's interesting.

03:25.165 --> 03:28.325
If that gene doesn't work properly, then you get diabetes.

03:29.315 --> 03:30.565
Does that tell us something helpful

03:31.255 --> 03:32.565
about what causes diabetes?

03:32.565 --> 03:36.885
And this approach has been applied to a lot of disease

03:36.885 --> 03:39.645
and genes and and you don't know whether you're gonna find

03:39.645 --> 03:41.165
out something helpful or not

03:41.705 --> 03:44.685
unless you do that to do those analysis.

03:45.225 --> 03:49.165
So the Biobank data, what, what people go in

03:49.225 --> 03:50.925
as they volunteer by main, they go

03:50.925 --> 03:52.165
through all sorts of assessments.

03:52.235 --> 03:53.485
They are, they go, one

03:53.485 --> 03:55.285
of the things they do is they get asked about their health,

03:55.355 --> 03:56.365
what diseases they have.

03:56.365 --> 03:57.965
They get asked what medication they're taking.

03:58.915 --> 04:03.485
Also, they, they consent for UK Biobank to examine the,

04:03.825 --> 04:06.765
health records that the National Health Service holds on

04:06.765 --> 04:09.245
people every time they go into hospital, if there's,

04:09.245 --> 04:11.605
there's discharge, there's diagnosis, they get assigned

04:11.825 --> 04:16.605
or, if, if, I mean people joined

04:17.155 --> 04:18.205
many years ago now.

04:18.205 --> 04:21.125
So if somebody's died in the meantime, then you get access

04:21.125 --> 04:24.085
to the health, code of what their course of death was.

04:24.825 --> 04:26.765
Um, and then people will do blood tests.

04:26.765 --> 04:27.965
They get all sorts of things measured.

04:28.585 --> 04:31.565
And some of 'em have been scanned an X-ray,

04:31.565 --> 04:32.925
an MRI scan and what have you.

04:33.145 --> 04:35.885
So the, the, the information that I used for this was,

04:35.925 --> 04:36.925
I wanted to say, I want

04:36.925 --> 04:38.765
to find out the people who've got type two diabetes

04:39.305 --> 04:40.805
and you've got 200,000 people there.

04:40.805 --> 04:42.485
Some of 'em have diabetes and some of 'em don't.

04:43.225 --> 04:45.325
And so I looked at a number of different things.

04:45.405 --> 04:48.165
I looked at in their original thing had they said,

04:48.715 --> 04:51.405
I've got type two diabetes as one of their health things.

04:52.005 --> 04:55.965
I looked at, have they got a ICD 10 code from, from a,

04:56.245 --> 04:58.245
a hospital appointment or commission

04:58.245 --> 05:00.445
or whatever saying they've got type two diabetes.

05:01.065 --> 05:02.565
And the other thing I did was I looked at the list

05:02.565 --> 05:03.925
of medications that they were taking

05:04.705 --> 05:07.885
and I said, have, are they taking a medication?

05:07.885 --> 05:10.045
And that medication is generally used

05:10.225 --> 05:12.045
for people with type two diabetes.

05:12.105 --> 05:13.125
And I took, so I took all

05:13.125 --> 05:14.165
these bits of information together.

05:14.165 --> 05:15.365
Now of course, each one

05:15.365 --> 05:18.725
of them itself might not be a hundred percent accurate, but,

05:18.945 --> 05:20.325
and you can certainly argue about that.

05:20.325 --> 05:22.405
There'll be some false negatives, some false positives.

05:22.405 --> 05:23.485
Some people there don't have diabetes,

05:23.485 --> 05:24.605
some people will be missed.

05:25.225 --> 05:28.125
But by and large, when you've got thousands, thousands

05:28.125 --> 05:31.205
of people, then you are getting to,

05:31.505 --> 05:32.685
you're getting a rough idea.

05:32.705 --> 05:34.805
You've got many, many thousand people who've probably got

05:34.805 --> 05:37.285
type two diabetes and you've got many other thousands

05:37.285 --> 05:39.445
of people who don't have type two diabetes.

05:40.025 --> 05:44.215
Uh, and so I simply took those as cases and controls.

05:44.815 --> 05:46.415
I looked at the exome sequence data.

05:46.535 --> 05:49.495
I looked to the, the sequence data for one gene at a time.

05:50.075 --> 05:51.415
And then I looked for each gene.

05:51.495 --> 05:55.575
I said, are there DNA variants which will damage that gene,

05:55.925 --> 05:57.805
will stop it working completely or,

05:58.025 --> 06:00.245
or will produce a different protein,

06:00.545 --> 06:02.365
an amino acid change that might be harmful.

06:02.785 --> 06:05.405
But those kind of things brought all that to, and,

06:05.505 --> 06:08.325
and I said, are the people for any given gene,

06:09.345 --> 06:12.245
are the people with type two diabetes more likely

06:12.265 --> 06:16.285
to have DNA variants which damage that gene than the people

06:16.285 --> 06:17.565
who don't have type two diabetes?

06:18.405 --> 06:22.045
I made a if for each gene, for each person, I made a, a sort

06:22.045 --> 06:25.285
of damage score, a burden score of the, the

06:25.805 --> 06:29.325
whatever DNA variants they carried, weighted by how damaging

06:29.325 --> 06:32.885
that variant would be to the gene, how,

06:33.315 --> 06:35.445
what the size effect it might be on a protein.

06:35.505 --> 06:36.965
So it might be that you didn't expect

06:36.965 --> 06:38.005
to make no protein at all,

06:38.185 --> 06:40.245
or it might be that you'd expect to make a protein,

06:40.305 --> 06:42.645
but it would be different and it wouldn't work very well.

06:43.135 --> 06:46.285
Added up that sort of damage score, the cost of variance

06:46.285 --> 06:47.525
for every gene for every person.

06:47.985 --> 06:49.965
And I said, have the people for every gene I said,

06:49.965 --> 06:53.405
have the people with type two diabetes got a higher weighted

06:53.405 --> 06:54.885
burden score than the people

06:54.885 --> 06:56.485
that don't have type two diabetes.

06:57.265 --> 07:00.245
And, and then you, you do that for 20,000 genes

07:00.245 --> 07:01.525
and you see which genes,

07:01.525 --> 07:03.565
have the highest scores for that test.

07:03.565 --> 07:04.685
There were three genes.

07:05.025 --> 07:07.565
But once you, you, you know, you're testing 20,000 genes.

07:07.565 --> 07:11.765
So you need a lot of evidence, very small p value,

07:12.545 --> 07:14.325
to say that this is statistically significant.

07:14.325 --> 07:16.965
There were three, three genes that passed that test.

07:17.585 --> 07:20.165
Two were already known, GCK

07:20.425 --> 07:23.005
and H NF four A.

07:23.865 --> 07:26.445
Um, and we say,

07:26.555 --> 07:28.405
I'll say something about those later, I expect.

07:28.745 --> 07:31.605
And then there was one gene which hadn't been identified

07:31.605 --> 07:32.765
as a risk factor for diabetes

07:32.765 --> 07:36.445
before, which was G giggy giggy, F1,

07:36.605 --> 07:39.125
GIGY F1.

07:39.395 --> 07:42.325
It's got its name is no more helpful than that.

07:42.325 --> 07:43.965
So I'll just call it Giggy F1 for now.

07:44.465 --> 07:45.725
Um, and,

07:45.985 --> 07:50.165
and my, I just picked, I just got in first

07:50.825 --> 07:53.725
as the first study to identify GG F1

07:53.945 --> 07:55.325
as a diabetes risk gene.

07:55.345 --> 07:56.765
And then very soon

07:56.765 --> 07:58.605
after another study was published,

07:58.605 --> 08:02.165
which said the same thing, and you know, pretty much my,

08:02.165 --> 08:03.325
I think mom was just in first,

08:03.325 --> 08:05.445
but they, I wouldn't wanna argue about it.

08:05.465 --> 08:08.205
And and that study actually had access

08:08.265 --> 08:10.365
to a bigger data set.

08:10.405 --> 08:13.045
I only had 200,000. They had three 50,000 I think.

08:13.385 --> 08:16.485
Uh, 'cause it was, that they were sort of internal,

08:16.585 --> 08:18.645
to, to develop doing the X sequencing.

08:18.945 --> 08:20.285
So, so those are the findings really.

08:20.285 --> 08:22.605
Three genes, two, three genes for type two diabetes.

08:22.805 --> 08:25.885
I, I guess again, I'll come back to this later,

08:26.085 --> 08:29.205
I think it's probably the most useful thing to come out is

08:29.205 --> 08:32.645
that the thing to I would want to emphasize is that very,

08:32.795 --> 08:36.605
very, very few people with type two diabetes have got one

08:36.605 --> 08:37.885
of these variants in one of these things.

08:37.885 --> 08:39.285
Fewer than 1%. I said.

08:39.285 --> 08:43.245
So only one in a hundred people with D 99%

08:43.245 --> 08:44.725
or more of people with type two diabetes.

08:45.185 --> 08:48.005
If you look at the whole exome, you will not find

08:48.805 --> 08:50.445
a genetic variant in a gene.

08:50.545 --> 08:52.765
Say, that's why they've got type two diabetes.

08:52.945 --> 08:54.685
And I think that probably

08:55.385 --> 09:00.285
Has the most important implications of, of anything

09:00.285 --> 09:01.485
that's sort of come from this study.

09:01.985 --> 09:03.685
My takeaway conclusion from this study,

09:04.185 --> 09:08.005
and I mean it's a big, big statement I think from this study

09:08.065 --> 09:11.365
and others that I have done on or that done myself

09:11.365 --> 09:14.085
and have been involved with these XM sequencing studies.

09:15.825 --> 09:19.565
My i is that, is that when somebody turns up at their GP

09:20.205 --> 09:23.735
practice with a disease, then essentially

09:25.245 --> 09:27.895
getting their DNA sequence and studying it

09:27.895 --> 09:29.255
and saying, ha you know, we are gonna

09:29.915 --> 09:31.615
do personalized medicine on you.

09:31.945 --> 09:33.895
We're gonna find out why you've got this disease.

09:34.225 --> 09:37.295
We're gonna get you, once we know that your genetic profile,

09:37.295 --> 09:39.575
we're gonna give you the treatment that will best suit you

09:39.815 --> 09:41.215
'cause of the genetic profile you've

09:41.215 --> 09:42.415
got now of that's gonna work.

09:42.435 --> 09:45.935
That's my, that's my conclusion, for disease

09:45.935 --> 09:47.375
after disease after disease that we've got had,

09:47.375 --> 09:48.975
including this one is one example of it.

09:49.675 --> 09:52.975
Almost nobody with the disease has got a genetic variant,

09:53.505 --> 09:56.095
which, you can say, they've got the disease

09:56.095 --> 09:57.095
because of that variant.

09:57.765 --> 10:01.015
What, you know, one in 200 people might,

10:01.115 --> 10:03.095
but the rest of them you've got no idea.

10:03.835 --> 10:05.855
And actually I'd even add that even if you do see

10:05.965 --> 10:08.575
that somebody's got a particular variant, you say,

10:08.715 --> 10:09.935
you've got type two diabetes

10:09.975 --> 10:13.015
because you've got a GCK mutation, you know you've got it

10:13.095 --> 10:14.935
'cause you've got a GigE F1 mutation.

10:15.525 --> 10:18.815
Even if you see that, does that mean you then know

10:18.815 --> 10:19.855
how to treat the disease?

10:19.855 --> 10:22.495
Does that helped at all? Is treatment any different answer?

10:22.995 --> 10:26.215
No, it's not. So, so what I think the conclusion is, is

10:26.215 --> 10:28.095
that this whole notion of personalized medicine

10:28.095 --> 10:30.775
that was sold to us, I mean, no, don't use

10:30.775 --> 10:33.695
that language sold to us, but was promoted, you know, 10,

10:33.695 --> 10:36.215
20 years ago with this, technology's gonna turn medicine

10:36.215 --> 10:37.295
around, transform medicine.

10:37.465 --> 10:39.015
It'll be the, the right treatment,

10:39.015 --> 10:40.295
the right patient at the right time.

10:40.755 --> 10:43.335
You go in, you have a blood test, you get thing,

10:43.355 --> 10:44.935
you get the right prescription based

10:44.935 --> 10:46.135
on this, that's not gonna happen.

10:46.795 --> 10:51.375
It do, doesn't work. Um, and that's kind of interesting.

10:51.515 --> 10:53.695
Now that's not the tool to say these studies don't have

10:53.695 --> 10:57.015
value, but they, that they don't have that value.

10:57.555 --> 10:59.895
The value that they have is what I said

10:59.895 --> 11:04.545
before is if you find a particular gene is abnormal

11:04.605 --> 11:06.465
and that kind of gives you the disease

11:06.465 --> 11:08.025
or massively increases the risk of disease,

11:08.325 --> 11:11.705
or conversely, if it's protected against the disease, then

11:11.705 --> 11:15.585
that can be helpful in terms of understanding the,

11:15.645 --> 11:17.545
the proteins which involve in molecules involved

11:17.685 --> 11:19.305
by chemical processes that are involved.

11:19.645 --> 11:21.865
And that might be helpful, you know, in the disease.

11:22.525 --> 11:24.225
And then that might be helpful in terms

11:24.225 --> 11:26.545
of developing new treatments and so on.

11:27.185 --> 11:29.325
Now, I, I would add that I don't think it has been helpful

11:29.545 --> 11:32.205
for type two diabetes and I can talk about

11:32.205 --> 11:33.645
what these genes do a little bit

11:33.645 --> 11:34.565
and it's kind of, it's

11:34.725 --> 11:35.885
interesting, but I'm not sure it's helpful.

11:36.505 --> 11:40.165
Um, but one doesn't doesn't know until one asks

11:40.815 --> 11:42.325
until one does that investigation.

11:42.985 --> 11:44.725
So, you know, it might've been

11:44.725 --> 11:47.685
that we found some completely unexpected gene,

11:47.735 --> 11:49.245
which did something that we never

11:49.245 --> 11:50.365
thought was anything through with diabetes.

11:50.385 --> 11:52.005
And we go, my goodness me, look at this.

11:52.425 --> 11:53.885
Uh, that's, that can give you diabetes.

11:53.965 --> 11:55.365
I wonder why that is. Let's develop this.

11:55.745 --> 11:57.845
You know, we've got new insight from that and so on.

11:58.225 --> 12:00.645
Uh, and I think that in this study that isn't really

12:01.315 --> 12:02.165
what has happened.

00:00 - There's this sample called the UK Biobank, which is half a million volunteers, in Britain who've given their health information and tissue samples and undergone various assessments. And one of the things they've done was to give a, a sample for the DNA

00:15 - and that had been the subjects of many, many studies called genome-wide association studies using, a million markers spread across all 23 chromosomes. of, and these are markers where there's common variation in people's genomes. Sometimes they have 1D NA based, sometimes they have another. And there've been lots and lots of these studies, done before these genetic studies. What was new was that rather than just,

00:44 - have these common markers, the UK bank bank had organized for these, this DNA to be what's called exome sequence, which means you read every single DNA base in every single gene. So you've got 20,000 genes. You read every single base in every one of those genes. And that really has, that differs from the previous Genome Wide association studies in, in two ways really. One is you are looking at variants that are actually

01:10 - inside genes that are the kind of important bits of the genome, the one or 2% which is actually coding for proteins and kind of really doing all the work. So you're looking in the important places, whereas genome-wide association studies, their markers are spread across all the chromosomes. A lot of 'em are not in genes at all, nowhere near genes. We dunno what they do. So you're looking at the genes. And the other thing is, is that

01:31 - because you are reading every single base in how many people, thousands and thousands of people, instead of just looking for common variation, you can see if somebody has a very, very rare variance that maybe only they have or mainly only half a dozen people have. So you find rare variance in genes. And the difference between rare variants and common variants is common variants. We all have lots of variation in code

01:58 - and it has almost no effect on, on us. It has tiny effects. And if you think about it, it's almost sort of common sense really with hindsight, that that would be the case. Because if anything had a very, if there's a change that had a very big effect, made us much taller, made us much smarter, whatever it was then, or made us much more sick, any big effect, you get evolution happening

02:22 - and it gets either, if it's advantageous, everybody gets it. And if it's disadvantageous, then nobody gets it. So the common variation that all around doesn't really do very much, but occasionally you can have a, a rare variant will occur and it will survive in the population for a while, even if it's quite harmful, even if it kind of really increases your risk of getting a disease such as diabetes. And so what had happened was you get back, biobank as said,

02:49 - organized for, it was at that time 200,000 exome sequences to be released from their volunteers. And this gave the opportunity to look at these 200,000 exomes and see if there are any very rare variants which might have very big effects in terms of increasing risk of, of diabetes or one of a number of other diseases. So this was what I was going into to find, and the i the point being that if you can then find that if there's a rare variant,

03:16 - which has a big effect in terms of damaging a gene and that gives you a disease like diabetes, then you can say, well, that's, you know, potentially you'll say, that's interesting. If that gene doesn't work properly, then you get diabetes. Does that tell us something helpful about what causes diabetes? And this approach has been applied to a lot of disease and genes and and you don't know whether you're gonna find out something helpful or not

03:41 - unless you do that to do those analysis. So the Biobank data, what, what people go in as they volunteer by main, they go through all sorts of assessments. They are, they go, one of the things they do is they get asked about their health, what diseases they have. They get asked what medication they're taking. Also, they, they consent for UK Biobank to examine the, health records that the National Health Service holds on

04:06 - people every time they go into hospital, if there's, there's discharge, there's diagnosis, they get assigned or, if, if, I mean people joined many years ago now. So if somebody's died in the meantime, then you get access to the health, code of what their course of death was. and then people will do blood tests. They get all sorts of things measured. And some of 'em have been scanned an X-ray, an MRI scan and what have you.

04:33 - So the, the, the information that I used for this was, I wanted to say, I want to find out the people who've got type two diabetes and you've got 200,000 people there. Some of 'em have diabetes and some of 'em don't. And so I looked at a number of different things. I looked at in their original thing had they said, I've got type two diabetes as one of their health things. I looked at, have they got a ICD 10 code from, from a, a hospital appointment or commission

04:58 - or whatever saying they've got type two diabetes. And the other thing I did was I looked at the list of medications that they were taking and I said, have, are they taking a medication? And that medication is generally used for people with type two diabetes. And I took, so I took all these bits of information together. Now of course, each one of them itself might not be a hundred percent accurate, but,

05:18 - and you can certainly argue about that. There'll be some false negatives, some false positives. Some people there don't have diabetes, some people will be missed. But by and large, when you've got thousands, thousands of people, then you are getting to, you're getting a rough idea. You've got many, many thousand people who've probably got type two diabetes and you've got many other thousands of people who don't have type two diabetes.

05:40 - and so I simply took those as cases and controls. I looked at the exome sequence data. I looked to the, the sequence data for one gene at a time. And then I looked for each gene. I said, are there DNA variants which will damage that gene, will stop it working completely or, or will produce a different protein, an amino acid change that might be harmful. But those kind of things brought all that to, and, and I said, are the people for any given gene,

06:09 - are the people with type two diabetes more likely to have DNA variants which damage that gene than the people who don't have type two diabetes? I made a if for each gene, for each person, I made a, a sort of damage score, a burden score of the, the whatever DNA variants they carried, weighted by how damaging that variant would be to the gene, how, what the size effect it might be on a protein. So it might be that you didn't expect to make no protein at all,

06:38 - or it might be that you'd expect to make a protein, but it would be different and it wouldn't work very well. Added up that sort of damage score, the cost of variance for every gene for every person. And I said, have the people for every gene I said, have the people with type two diabetes got a higher weighted burden score than the people that don't have type two diabetes. And, and then you, you do that for 20,000 genes and you see which genes,

07:01 - have the highest scores for that test. There were three genes. But once you, you, you know, you're testing 20,000 genes. So you need a lot of evidence, very small p value, to say that this is statistically significant. There were three, three genes that passed that test. Two were already known, GCK and H NF four A. and we say, I'll say something about those later, I expect.

07:28 - And then there was one gene which hadn't been identified as a risk factor for diabetes before, which was G giggy giggy, F1, GIGY F1. It's got its name is no more helpful than that. So I'll just call it Giggy F1 for now. and, and my, I just picked, I just got in first as the first study to identify GG F1 as a diabetes risk gene.

07:55 - And then very soon after another study was published, which said the same thing, and you know, pretty much my, I think mom was just in first, but they, I wouldn't wanna argue about it. And and that study actually had access to a bigger data set. I only had 200,000. They had three 50,000 I think. 'cause it was, that they were sort of internal, to, to develop doing the X sequencing.

08:18 - So, so those are the findings really. Three genes, two, three genes for type two diabetes. I, I guess again, I'll come back to this later, I think it's probably the most useful thing to come out is that the thing to I would want to emphasize is that very, very, very few people with type two diabetes have got one of these variants in one of these things. Fewer than 1%. I said. So only one in a hundred people with D 99% or more of people with type two diabetes.

08:45 - If you look at the whole exome, you will not find a genetic variant in a gene. Say, that's why they've got type two diabetes. And I think that probably Has the most important implications of, of anything that's sort of come from this study. My takeaway conclusion from this study, and I mean it's a big, big statement I think from this study and others that I have done on or that done myself and have been involved with these XM sequencing studies.

09:15 - My i is that, is that when somebody turns up at their GP practice with a disease, then essentially getting their DNA sequence and studying it and saying, ha you know, we are gonna do personalized medicine on you. We're gonna find out why you've got this disease. We're gonna get you, once we know that your genetic profile, we're gonna give you the treatment that will best suit you 'cause of the genetic profile you've got now of that's gonna work.

09:42 - That's my, that's my conclusion, for disease after disease after disease that we've got had, including this one is one example of it. Almost nobody with the disease has got a genetic variant, which, you can say, they've got the disease because of that variant. What, you know, one in 200 people might, but the rest of them you've got no idea. And actually I'd even add that even if you do see that somebody's got a particular variant, you say,

10:08 - you've got type two diabetes because you've got a GCK mutation, you know you've got it 'cause you've got a GigE F1 mutation. Even if you see that, does that mean you then know how to treat the disease? Does that helped at all? Is treatment any different answer? No, it's not. So, so what I think the conclusion is, is that this whole notion of personalized medicine that was sold to us, I mean, no, don't use that language sold to us, but was promoted, you know, 10,

10:33 - 20 years ago with this, technology's gonna turn medicine around, transform medicine. It'll be the, the right treatment, the right patient at the right time. You go in, you have a blood test, you get thing, you get the right prescription based on this, that's not gonna happen. It do, doesn't work. and that's kind of interesting. Now that's not the tool to say these studies don't have value, but they, that they don't have that value.

10:57 - The value that they have is what I said before is if you find a particular gene is abnormal and that kind of gives you the disease or massively increases the risk of disease, or conversely, if it's protected against the disease, then that can be helpful in terms of understanding the, the proteins which involve in molecules involved by chemical processes that are involved. And that might be helpful, you know, in the disease. And then that might be helpful in terms

11:24 - of developing new treatments and so on. Now, I, I would add that I don't think it has been helpful for type two diabetes and I can talk about what these genes do a little bit and it's kind of, it's interesting, but I'm not sure it's helpful. but one doesn't doesn't know until one asks until one does that investigation. So, you know, it might've been that we found some completely unexpected gene,

11:47 - which did something that we never thought was anything through with diabetes. And we go, my goodness me, look at this. that's, that can give you diabetes. I wonder why that is. Let's develop this. You know, we've got new insight from that and so on. and I think that in this study that isn't really what has happened.

Faculti AI can make mistakes