Host: Benjamin Thompson
Welcome back to the Nature Podcast. This week, using deep learning for a better weather forecast…
Host: Shamini Bundell
And an atlas of the mammalian motor cortex. I’m Shamini Bundell.
Host: Benjamin Thompson
And I’m Benjamin Thompson.
[Jingle]
Host: Benjamin Thompson
First up this week, reporter Ali Jennings learns about a powerful new technique for that great British pastime – predicting the rain.
Interviewer: Ali Jennings
It’s the age-old question: should I have brought my umbrella today? Predicting precipitation in the very near future is important in many situations – managing emergency services, giving early flood warnings and controlling air traffic, as well, of course, as optimising your daily commute. This kind of short-term forecasting is known as nowcasting – predicting precipitation patterns up to just two hours ahead. One popular way to do this is to use radar. Microwave pulses bounce back off airborne particles and can be measured to provide a picture of the surrounding precipitation. By combining radar data with an atmospheric physics equation that calculates how air moves, nowcasters can build up short-term precipitation predictions. But there’s a problem.
Interviewee: Suman Ravuri
Yeah, so that method is not particularly great at predicting thunderstorms and other heavy precipitation events.
Interviewer: Ali Jennings
This is Suman Ravuri from the company DeepMind in the UK. Suman is part of a team exploring how deep learning – a particular kind of artificial intelligence – could help solve nowcasting problems like these. The physics equations used in nowcasting rely on assumptions that don’t always map onto reality, but deep learning models don’t rely on these equations. Instead, they’re trained entirely on radar data, and lots of it. For example, Suman’s model is trained on 1.5 million examples of previous rainfall in the UK. Suman and the DeepMind team aren’t the first to use deep learning for nowcasting, but their model is a little different from what has come before. It’s called a deep generative model.
Interviewee: Suman Ravuri
So, a deep generative model is a model that gives you alternate futures, in this case, of precipitation.
Interviewer: Ali Jennings
Suman and his team trained their new deep generative model to imagine what future rain might look like, based on radar data taken over the previous 20 minutes.
Interviewee: Suman Ravuri
We spent a lot of time making sure that the predictions themselves didn't only look like pretty videos of rain but were ones that were actually consistent with what happened in reality. It's really exciting kind of seeing this model as it’s training over time, going from making predictions that look nothing like reality to sort of slowly getting to things that look like possible evolution of precipitation.
Interviewer: Ali Jennings
The predictions made by the DeepMind team’s model are judged by another program called a discriminator. The discriminator tries to discriminate between the precipitation predictions the nowcasting model has generated and what actually happened in reality. When their discriminator couldn’t tell the difference between their model and reality, the team were happy with their model’s accuracy, and they asked 56 expert meteorologists from the UK Met Office to judge the predictions of their model against predictions from other existing models.
Interviewee: Suman Ravuri
They preferred our predictions 89% of the time. And they also told us that this is something that is a step change in terms of what they're used to working with.
Interviewer: Ali Jennings
The model had much higher detail than what meteorologists are used to and was the best at representing risk, according to some of the meteorologists who took part. And Suman wants his model to have real, practical applications.
Interviewee: Suman Ravuri
I hope either this or some future work of this does form the base of predictions for issuing flood forecasting or other warnings.
Interviewee: Suzanna Maria Bonnet
In Rio de Janeiro, some streets have flood problems and landslides, and it's very important to take decisions to mobilise security sectors.
Interviewer: Ali Jennings
This is Suzanna Maria Bonnet, a forecaster and researcher at the Federal University of Rio de Janeiro.
Interviewee: Suzanna Maria Bonnet
This type of model is very important to give this type of information to the meteorologists who will call the authorities.
Interviewer: Ali Jennings
But although Suzanna thinks that DeepMind’s model could be extremely useful for disaster prediction, she points out a significant problem for making this model accessible to a wide group.
Interviewee: Suzanna Maria Bonnet
We can have a very big, complex model, but if we don’t have computational resource to run this model, we will have to cut the model.
Interviewer: Ali Jennings
And there’s a more specific issue that could affect whether forecasters like Suzanna could use this prediction tool, this time with the parameters within the model shaping its predictions.
Interviewee: Suzanna Maria Bonnet
If I want to use this model here in Brazil, some parameters are not specific for my region. it's a problem we have with numerical weather prediction models. Many studies that change parameterisations inside the model, they are made in the higher latitudes. It's not applied for our tropical region. It changes a lot the results.
Interviewer: Ali Jennings
Even back in the UK, Suman appreciates that they’ll need to work on the model further with meteorologists and experts in the field to get it to a stage where it can be deployed in practice. But he and Suzanna agree that they’d also like the model to take account of more different kinds of data.
Interviewee: Suman Ravuri
I think maybe we’re, at the moment, 20 to 30% of the way there. I think there’s a lot more work that we can do about incorporating other types of weather data. However, what’s really interesting to me is incorporating more physics into our deep learning models. So, that’s something that we haven’t scratched the surface of yet and something that I really hope that a lot of people think about and work on.
Interviewer: Ali Jennings
And predicting rain isn’t the only thing on DeepMind’s radar. Wind patterns and temperature change are both things DeepMind are considering tackling in the future. So, perhaps one day I’ll finally know what to wear on the way to work.
Host: Benjamin Thompson
That was Ali Jennings. For this story he spoke to Suman Ravuri from DeepMind in the UK and Suzanna Maria Bonnet from the Federal University of Rio de Janeiro in Brazil. To find out more, look for a link to the paper in the show notes.
Host: Shamini Bundell
Coming up, we’ll be hearing about a huge undertaking to map the mammalian brain, starting with the motor cortex. And later on, Flora Graham from the Nature Briefing will be joining us to discuss the winners of this year’s science Nobels. Now, though, Dan Fox is here with this week’s Research Highlights.
[Jingle]
Dan Fox
Vaping has been considered by many to be a safer alternative to smoking cigarettes. But despite this, vaping devices, which can produce a nicotine-laced vapour, have been linked to more than 2,500 hospitalisations and dozens of deaths in the United States alone. Vaping’s worst health effects seem to be linked to a compound called vitamin E acetate, which can be added to vaping liquid, but how the additive damages the lungs is unclear. Now, a team of researchers have devised a way to try to understand its effects – a vaping robot. The team’s tabletop device can inhale much as a human does. The robot contains air heated to body temperature and a laser-based sensor that measures how many particles of various sizes are pulled in during ‘inhalation’. Initial experiments suggest that vitamin E acetate enhances the number of particles smaller than 10 micrometres that are inhaled from the vaping device. Such small particles could go on to lodge in the lungs, fostering or exacerbating damage. Read that research in full in iScience.
[Jingle]
Dan Fox
Some sea slugs might be more fertile when the Sun shines thanks to actively photosynthesising organelles taken from seaweed. Sacoglossans are a superorder of marine molluscs who feed on the cellular contents of seaweed and algae. Some of the slugs have managed to incorporate the algae’s photosynthetic structures, called chloroplasts, into their own cells, using them to obtain nutrients from sunshine. One slug species, Elysia timida, even moves these chloroplasts into its reproductive tissues, and now a team of researchers have observed that the slugs produce more eggs when in the presence of light than when they’re in darkness. The team show the stolen chloroplasts produce molecules the slugs use to build polyunsaturated fatty acids important for reproduction. Go out in the sunshine and read that paper in full in The Proceedings of the Royal Society B.
[Jingle]
Host: Shamini Bundell
Back in 2013, the then US President Barack Obama announced the BRAIN Initiative, an ambitious project to revolutionise our understanding of the human brain, to do for neuroscience what the Human Genome Project did for genomics, and try to determine the role of different parts of the brain in disease. As part of this initiative, in 2017, a group of researchers formed to create a comprehensive map of the mammalian brain at a cellular level. Fast forward to this week, and this group has published 17 papers in Nature and more elsewhere. Now, to put it mildly, the brain is a complicated organ, so the researchers involved started by just mapping the cells in a small part – the motor cortex – and they’ve been doing so in several mammalian species. Reporter Nick Petrić Howe caught up with one of the researchers involved, Hongkui Zeng, to find out more. He started off by asking, why the motor cortex?
Interviewee: Hongkui Zeng
The motor cortex is a well conserved region across different species, so the conservation with allow us to be able to reliably identify that region in, for example, an animal, a mouse brain or a human brain. And then as a starting point, we can use that conserved region to see if the cellular composition is also conserved or not across species.
Interviewer: Nick Petrić Howe
Even though it’s a smaller region in the brain, it still seems like a huge undertaking. How has it gone, trying to map this region?
Interviewee: Hongkui Zeng
Yeah, the major challenge is to combine the different technologies together. If you just use one type of technique to profile the cells in the motor cortex, it can be done now, routinely, in different labs. However, if we use different techniques to get the information from the cells of that region, the result is telling us only one aspect of the cells. In order to see the whole picture, you really need to integrate different kinds of information together. So, I think that is where the challenge lies. And we use various computational approaches that allows us to integrate different kinds of information together.
Interviewer: Nick Petrić Howe
And, I mean, there are 17 papers being published in Nature this week and some more papers in the broader Nature family of journals, so there’s a lot of information here, but would you be able to share with me just some highlights of the sorts of things that you found through this project?
Interviewee: Hongkui Zeng
So, this set of papers really sets the foundation and a layout of the roadmap that will allow us to really, in the end, reach this comprehensive map of the entire brain in the not too distant future. So, that’s the first aspect of it. And then secondly, back to the motor cortex itself, we identified several major principles for the organisation of cell types in that region. The first is that the cell types are organised in a hierarchical manner. At a higher level, the distinction between different cell types is very robust and very clear, but when you go down to the more refined branches of the hierarchical tree, the distinction between the more detailed types becomes fuzzier, and we think that this is really a general organisation of the cells across the entire brain. The first discovery is the organisation of the cells. The second is then the mechanism behind that organisation. The gene expression and the genomic landscape define the unique properties of the different cell types and the major distinctions between different cell types. Another major discovery is that we find those major cell types that we defined in this way are well conserved across species, from the mouse brain to the human brain.
Interviewer: Nick Petrić Howe
So, this hierarchy of cells are the different groups of cell types. So, for example, there are inhibitory neurons and non-inhibitory neurons, and they’re very distinct. But then as you look at the different subtypes within that, things get a bit fuzzier. Was this something that you were expecting?
Interviewee: Hongkui Zeng
This hierarchical structure has already been recognised in previous studies. The major finding of this study is then that we are able to combine the different types of properties together, not just the gene expression but also the shapes of the cells, the physiological properties of the cells, combining all of them together and we see that even by combining the different properties together, the cells still follow the same kind of hierarchical organisation.
Interviewer: Nick Petrić Howe
So, what do you think will be the impact on the field of these findings?
Interviewee: Hongkui Zeng
I think the major impact is that this set of studies really sets a roadmap of how to develop an atlas or map of the cells across the entire brain. And then going from here it’s pretty straightforward. We can use the same approach to profile cells in other regions and generate a much more comprehensive map.
Interviewer: Nick Petrić Howe
I imagine I know what the answer to this question is but does this close the book on the motor cortex, or is there a lot more work to be done?
Interviewee: Hongkui Zeng
I think in terms of the cellular composition of the motor cortex, it pretty much closes the book on this region. What should be done next? What needs to be done next even for the motor cortex is to first understand how the cells are interacting with each other. At the most basic level, it’s to understand how the cells are connected to each other at a synaptic level. Synaptic connections are really the cardinal features of the brain, of the nervous system, how cells communicate with each other and how they generate thoughts, emotions, flexible behaviours and things like that.
Interviewer: Nick Petrić Howe
And one thing I wanted to ask you about, and it’s of growing interest for many countries and governments, is there are ageing populations, there are a lot of neurological conditions, Alzheimer’s, for instance. How could this work help in this sort of area with therapeutics?
Interviewee: Hongkui Zeng
What we have now is a foundational reference map for the motor cortex as an example, and very soon we will have a reference map of cells across the entire brain, as we were just talking about. And that really sets the stage and is a starting point for them to see how the brain changes in a variety of different diseased conditions, and then pinpoint exactly where the changes are happening and what kind of genes whose expressions are changed in what kind of cells. I believe this knowledge will be tremendously informative in telling us what has gone wrong in a diseased brain and what kind of targets we might use to treat the diseases.
Host: Shamini Bundell
That was Hongkui Zeng from the Allen Institute for Brain Science in Seattle, Washington in the US. To find out more about this research, check out the links in the show notes.
Host: Benjamin Thompson
It’s Nobels week this week so, as has been become something of a tradition around these parts, Flora Graham, senior editor of the Nature Briefing, joins me to discuss this year’s winners. Flora, how are you doing today?
Host: Flora Graham
I’m doing great. It’s great to be here. Thank you so much for having me.
Host: Benjamin Thompson
So, I think this is the fifth year in a row we’ve done this, you and I.
Host: Flora Graham
Wow, truly, it’s my turn to win a Nobel soon then.
Host: Benjamin Thompson
Yet again, sadly, you have been overlooked this time round. But let’s go through who has won and let’s start on Monday with the Nobel Prize in Physiology or Medicine. Who are the winners this year?
Host: Flora Graham
This year, the medicine prize went to a physiologist, David Julius, and a molecular neurobiologist, Ardem Patapoutian.
Host: Benjamin Thompson
And I think they’ve won really for helping us understand how we get a sense of the world, I suppose, through sensing heat and through touch.
Host: Flora Graham
And as Patapoutian said in an interview with the Nobel Prize committee, it’s really one of the most simple things in life. It’s one of the most obvious things in life. How do we touch? How does this sense work? And it’s the fact that they have both been able to find underlying mechanisms of something that all of us can appreciate is what makes this research so interesting, I think, to all of us.
Host: Benjamin Thompson
Well, let’s start with David Julius and what he won his half of the prize for then, and it kind of centres around capsaicin, which is the molecule which gives chilli it’s heat.
Host: Flora Graham
That’s right. This is a molecule that hot sauce fans will be very familiar with. This is the chemical that gives that little bit of heat, or some of us might consider it slightly painful, to spicy food, and what Julius and his team did was they tracked down a protein called TRPV1, and that’s the protein on the surface of the cell that actually responds to painful heat.
Host: Benjamin Thompson
And these kinds of cell surface proteins are really, really important to Patapoutian’s finding as well, but his work isn’t necessarily about heat, it’s about being pressed, mechanical forces.
Host: Flora Graham
Yeah, in this case, Patapoutian and his team identified cells that emitted an electrical signal when they were prodded, and that led to the discovery of ion channels, two ion channels named Piezo1 and Piezo2, which are activated by pressure.
Host: Benjamin Thompson
And as I understand, independently, they also kind of worked out how cells respond to cooling as well, and this, I guess, opens up a whole world of potential medical interventions into things that target pain or what have you.
Host: Flora Graham
That’s right. In addition to explaining this very fundamental, basic biology of the senses, these could have potential medical applications for, as you mentioned, combatting chronic pain. We have to look for compounds that target some of these proteins that Julius and Patapoutian discovered.
Host: Benjamin Thompson
Well, Flora, for listeners who have been following our chats about the Nobel winners over the past few years, we always end up talking about how one of the winners found out about their prize, and of course the time zones can be very much against them, and that certainly was the case for David Julius.
Host: Flora Graham
Absolutely, he told the Nobel Prize committee that it took ages for them to get in touch with them. So many of these scientists might be on the other side of the world when the announcements are made, and in his case, I think it ended up being his sister-in-law that finally broke through. I think the moral for us all is just leave your phones on on Nobel Prize announcement night. You never know, it might be you.
Host: Benjamin Thompson
Well, let’s move on to Tuesday, Flora, and the physics prize. Now, this was split between three researchers. Who are they?
Host: Flora Graham
Yeah, it was quite an interesting winning group for physics this year. The prize is shared between Syukuro Manabe, Klaus Hasselmann and Giorgio Parisi. Now, the first two, very strong in the field of climatology, and Giorgio Parisi, an incredibly wide-ranging physicist who has worked on many different areas but specifically in this case they referred to the complex systems research, specifically things like spin glass.
Host: Benjamin Thompson
Yeah, I think understanding complex systems seems to be quite at the heart of this prize, and of course with COP26 coming up, climate is very, very much on the agenda, and Manabe and Hasselmann’s work seems really, really important to the field.
Host: Flora Graham
Well, Göran Hansson, who’s the secretary-general of the Academy of Sciences said that the modelling underpinning our knowledge of climate change is solidly based on physical theory and solid physics, and that global warming and the theory behind it is resting on solid science.
Host: Benjamin Thompson
And this is work that’s been around for quite some time.
Host: Flora Graham
That’s right. Manabe’s research was among one of the very early climate models. These are models that were being built in the 60s and 70s, and they have proven to be very successful at predicting global warming. And Klaus Hasselmann wrote a paper that’s widely regarding as the first serious effort to provide a strong statistical framework for identifying a human-caused warming signal.
Host: Benjamin Thompson
Giorgio Parisi then, his work seems quite different at face value, but of course it is linked to these complex systems like the climate.
Host: Flora Graham
It is linked, and some people have said that this is maybe a little bit of a tenuous link, but the underlying theme is that complex systems, which might seem on the surface too complex to understand, actually, theorists are able to zoom out and create models that are very effective at understanding what’s going on holistically across the board.
Host: Benjamin Thompson
Let’s round it off then, Flora, and the prize for chemistry, which was announced today, just a few short hours ago. Who are the winners this time?
Host: Flora Graham
Well, Benjamin List and David MacMillan shared this year’s chemistry prize, and both of them did independent work in a technique called asymmetric organocatalysis, which is widely used today for production of drugs and other chemicals as well.
Host: Benjamin Thompson
Well, a super important technique in the field of chemistry, Flora, and developed in the early 2000s, as you say, independently, by these two researchers. What do we know about organocatalysis?
Host: Flora Graham
Well, the process relies on small, organic molecules rather than big, biological enzymes or compounds based on heavy metals. So, that can make it cheaper and it can be more environmentally friendly for some chemical reactions.
Host: Benjamin Thompson
And of course, we can’t overlook the ‘asymmetric’ aspect of asymmetric organocatalysis. This is really important for the process of making drugs and what have you.
Host: Flora Graham
That’s right. That’s where a reaction produces more of one version of a molecule than the other, so let’s say more left-handed versions of a molecule than right-handed versions. And this is important in medicine, as you mention, because two mirror images of a molecule can actually trigger very different biological effects.
Host: Benjamin Thompson
And what are people saying about the technique being at the heart of this year’s win?
Host: Flora Graham
Chemists are saying that this work really led to an explosion in the field, so it’s the kind of research that has had a huge knock-on effect. As the president of the American Chemical Society, H. N. Cheng, said, chemists are like magicians, and with this technique, it’s a new magic wand that they can use for making important drugs.
Host: Benjamin Thompson
Well, thank you for that roundup. One thing, Flora, that stands out, once again, is the lack of diversity in the winners this year.
Host: Flora Graham
Yeah, the diversity of the winners is something that the Nobel committee has pledged to address for several years now, specifically the gender balance is something that has just not stood up to the representation in science as a wider field, and yet it doesn’t look like the committee’s efforts to increase diversity have really come to pass this year.
Host: Benjamin Thompson
Well, Flora, thank you so much for joining me, and we’ll see you back here in a year’s time for the 2022 winners. And listeners, for even more Nobel updates, head over to nature.com/news. That’s about all we’ve got time for this week. But before we go, a quick plug for ‘Starting up in science’. It’s a podcast series that myself and a bunch of our colleagues here at Nature made in which we follow the trials and tribulations of two researchers over three years, as they struggled to get their labs off the ground. It’s a story that’ll be achingly familiar to many and, well, it goes some places. But rather than me spoil it here, look out for it wherever you get your podcasts.
Host: Shamini Bundell
We’ll be back next week, but don’t forget, in the meantime, you can drop us a line and say hi, either on Twitter - @NaturePodcast – or email – podcast@nature.com. I’m Shamini Bundell.
Host: Benjamin Thompson
And I’m Benjamin Thompson. Thanks for listening.