Episode 65: Making Connections – linking data on biodiversity

Naturally speaking. Hello and welcome to Naturally Speaking. I'm Tae Afford. And I'm Karsten Gwenny.

And we're joined today by Institute researcher Rod Page. His research is focused on evolutionary biology, phylogenetics, and bioinformatics. So Rod recently won joint first prize in the 2018 Ebby Nielsen Challenge, which is an annual innovation competition run by the Global Biodiversity Informatics facility. This is an international network and research infrastructure funded by the world's governments, and is aimed at providing anyone, anywhere, open access to data about all types of life on Earth.

So Professor Page, welcome to Naturally Speaking. Congratulations on this recent prize. Can you perhaps tell us a little bit more about the field of biodiversity informatics? I'm pleased to ask me to define it, because it's actually really hard to define.

I think it's essentially a intersection between biodiversity and computers. So probably the easiest way to explain it is if you go to the Jibber website, Jibber.org, what you'll see is a gigantic map with hundreds of millions of little yellow dots on the map. And those dots represent a place where somebody has said a species occurs at this place at this time. So that's one of the major things that we're doing.

We're going to find out what are these species, what they look like, who they're related to, where are they, and what are they doing. And it can be animals plants, fungi, bacteria, dinosaurs, anything that is alive or has been alive. Where is it, and what is it? It's quite an ambition, it is very ambitious.

So I guess I'm going to practice what that means as people are trying to literally go by obviously on that. Jibber itself, the organisation that had this competition started out by talking to Metro museums, like the one over here, Hyperia, and they have all these amazing collections of specimens, and often they'll be a little off and handle it and label it's saying what it is and where it was found. We want to convert that into a digital form, put it on database, put it online so everybody can play with it. And that's sort of been going for about maybe 15 years now.

So there's all the data from museums in Hyperia, there's stuff from citizen science, which has become a huge thing. We now have people wondering how their iPhones, taking pictures, they've got GPS, you know, exactly where it is. It's actually a huge time for the data that's now coming from just regular people, observing plants and animals and putting it online. So I guess Biodeversity is all about getting that fundamental information in a digital form and putting it online so people can play with it.

So your prize work was entitled Aussie Mandius, a Biodeversity Knowledge Graph. So first question, what is Ali as a Knowledge Graph, and also how does it help us in documenting or understanding biodiversity? That's a really good question. So a Knowledge Graph, it's a kind of marketing phrase that Google came up with for a way of connecting bits of information.

I was trying to think of a way to explain this. Many of us probably use databases. So if you are writing a paper, because these are something that you probably have something like in notes or mint layers of material to manage your references. And that's kind of a database, you've got some information and you've got this kind of list of papers that you can cite.

For a Knowledge Graph, they're not really interested in the paper so much, but the connections. So what you really want to know is which paper sites which other paper can you get links in terms of citation. If you're looking at authors, you're less interested in the particular author than if somebody often co-authors or somebody else. So you're looking for those sort of social connections.

And you can think of that, maybe that's the way. If you come to a field, you know nothing. One way you start with the paper and you see who the authors are, you see what they've written, you see what they've cited, and away you go. So Knowledge Graph is that kind of sense of these connections.

You're interested in linking things together. The other big thing that makes Knowledge Graph so that different is if I had a say database of papers and you had one, it's possible that we would store them in certain ways. I might have a long name, the journal, you might have a short one. So we had to read cool notes on that paper and I said notes, we want to get a motion together.

We're going to have to figure out which paper is the same one. So the thing that sort of rescues us in that situation is this idea of an identifier. You've probably seen these DOIs, but now in those papers. So if I have the DOI and you have the DOI, then we can say the same thing, we can look at it together.

So the big part about Knowledge Graph is you want the connections and to make the connections you want everything to have a nice unique identifier. And if you have those, you make the connections and away you go. So I guess the dream is that everything would be linked together. So if you're in some species, you could find what paper first described as species.

Show me the description. What kind of read about the ecology of the species from that? Where are these specimens? What museum are they?

Has anybody sequenced it? Is there a solution to tree for that? What's the ecology like and so on? I guess the idea is you sort of touch that one part of it and it worked in your way.

And the challenge is we don't have that. So the goal is to try and build it. So the thing with Ozymendius was even I'm not foolish enough to think we can build this with the entire planet one. Go ahead, that's just insane.

But for small sections of what we know, we could probably make a good go with it. And Australia had been some amazing stuff in the next few years. They put huge misinformation on their biodiversity online. And so I just sort of came along, grabbed that and said, it's really cool if you took that data and connect it together.

And that was my entry into this competition. You've mentioned a lot of different types of data out there. You've mentioned these museum collections. You've mentioned particularly citizen science.

And that's another way of tapping into a maybe a previously underused resource. So how do you go about finding all of these data sources? And especially in the case of citizen science, ensuring that we have data quality that we can actually rely on those data to be incorporated? So kind of a data, it depends against what species you're looking at and what country you're looking at.

So something like Australia, for example, they have been putting huge amount of effort into getting the data available. You find the websites. Some websites are really nice. There's a button saying, please download my data.

The websites are not so nice. And so you play various games about how you try and get the data out of them. There's all sorts of black arts that people use to get data. What you find is that data is not really suitable for use.

And you end up seeing a lot of time missing about with it. Quality is a huge issue. A lot of the data that we use comes from museums. And I guess the idea would be that museums would be a familiar source of information.

If you've ever been to museums and have looked behind the scenes, you'll often find sometimes the labels really old. Sometimes they're quite vague. And in many cases, the material hasn't been looked at by a text on us for a long time. So we have a lot of information online that could be quite seriously wrong.

There have been some studies where people have gone and looked at how many of the names that our museum collections have still been used. For example, if you wander around out one of the museums here, I get a second-year students to do that, and I get a number of emails when they do a project on that saying, what is this thing? And I have to explain that the 1970s was called that, now it's something else. And that becomes an issue.

There are all kinds of data quality issues. If you ever look closely at this amazing cheaper map, you'll find some weird things, and there are lots of observations that I can put at 45 degree angle. That's really unusual. Or there are straight lines.

And some of those are collecting lines, but some of those are simply mistakes in the data. For a long period, the most diverse part of the planet, Kune Jibov, was just off the coast of Africa, at zero degrees, latitude, zero degrees, longitude. Because people in South Square, didn't know what the point was, and it got interpreted as an actual place. And there's all sorts of way to multiple things.

It's very easy for people to swap latitude and longitude. There's a beautiful blog post by the people who work at Jibov, sharing your distribution of the world of everything that's recorded from the United States. And in the United States was the egg glowing, and then, outside the world, China was glowing. And if you look closely, you can actually see a mirror image of the United States.

So longitude and so being west was east, everything's in China. Because part of Australia was glowing, just a little bit. That's when people have such numbers around. So there are issues.

And people would be frustrated by that. Anybody who uses Jibov data, if you ever read any of the papers, they'll say we download the data from Jibov, and then we cleaned it. And they have to sort of detect things like old text, not names, places that just don't make any sense, and so on. So that's a challenge.

Everybody's kind of a weir of it, and nobody's come up with a super good way of doing it. So in some ways, I mean, it's citizen science. On the one hand, it's often people hand-drained, but on the other hand, if you go to the museums, some of the stuff is just so old and so poorly documented. You could argue that today you got the iPhone, you got a very precise latitude and longitude, more so than most previous observations.

You got a camera that goes and gets. There's an app that I use a lot called Tynetros, which can now tell you what the species is that you're looking at. Sort of like the point you're phoned at the stars, and it tells the constellation you've put on the line. So yeah, it's a constellation thing works because it knows where you are, and where you're pointing to.

The way that an interesting works is, I was in Germany over the summer, and I was walking on the side of the river. I took a picture of a green leaf with some red things dangling below it. I had no idea what it was. So I put on an actualist and this little button which goes to suggest.

So I did that, and it came back with the red nail might. And it turns out that's exactly what it is. So the inside of his leaves, he said, I mean, you know what, you're a petite. It's quite a common thing, but it's might.

And it's like these titles that looked like things. Nobody believes. I had no idea what that was. And it came back and said, it's a might.

You kind of rely on people taking photographs. So the kind of things that works out really well. Like you go on a campus here and photograph pretty much any tree or any plant in the garden. It'll figure out what it is because they're really, really common.

But often it would make a reasonable guess about what it is. So the way I actually try to ensure quality is I can look at the graph of a hover fly and upload it. Eventually that'll end up in G-Biff. But only if two or three other people come along and say, yes, it's a hover fly and we agree it's the same species.

Though if I just say it's a fly, I'm not sure what it is, that doesn't get through the process. So they've kind of got a sort of crowdsourcing identification thing going on. I love the same lines, and if you had any ideas about creating an app to do any of the bio-diggers or stuff that you're working on. I guess the idea of being able to identify something just by looking at it would be the thing.

It's a much simpler kind of thing. I mean for example, a copyist go on another project here with the idea was how do people find information in museums? So what we did is we made this with QR codes and we plastered them around the spike end of this museum store. And we then set up a little website.

And the idea was, I guess the sort of site equation behind it was if somebody goes to the museum, do they start at the front and sort of walk around the loop? Or do they head towards Gavin the Oryx? Or do they go for particular kind of species or areas? I was kind of like, we tried to picture it as an optimal foraging experiment.

And you could kind of figure out what people are doing in the museum and then what species are popular and so on. I guess the whole thing about apps is kind of interesting because for somebody who spends time developing websites and stuff, there's always a tension between do you make an app which is kind of cool but quite time consuming, and then if you make an app, do you have an iPhone or do you have an Android? Or can you make a website that is good enough on your phone? So for example, Jeeviv has gone down the roof.

We all have a website that works on your phone, works on your iPad, works on your computer. It's all the same thing. And hopefully that's good enough for what you want to do. Where do you see the field of biodiversity and informatics moving in the coming years and what are the biggest gaps that will be filled?

So one of the big gaps I guess is the classic geographical gap. So if you look at the Steve of Picture, the sort of map that they give you, it's gorgeous, but the brightest spots are Europe, North America, South Africa, Australia, Japan. So it's where people have time and money to get involved in digitizing information. The tropics is very, very sparse.

And Russia doesn't exist. And there's all kinds of issues. So there's huge gaps in our knowledge of where biodiversity is in terms of people just making it available digitally. So there's lots of interest in doing that.

And you have all sorts of challenges. And some of the people on the institute were in Rwanda earlier this year looking at some of these things. And it's just a different world in terms of kind of resources that they've got. Whereas here people go into museums and make really cool treaty models and that's what it's really hard to do.

So there's big challenges with that. There's big challenges in how you perceive biodiversity. So if you watch any of David Ettembert's wonderful programs, they're all basically very off-fetherely. That's how we perceive biodiversity, likewise in museums.

Whereas the Volcker biodiversity, of course, isn't that. So we have a very sort of disordered new biodiversity. And the other thing that there's a huge challenge for us is we are still in some ways in 19th century science in that sequins later is only slowly sort of coming into this kind of the organist sense. And the whole idea of the school stuff in DNA back-coding is metagenomics and environmental DNA.

And that stuff is not really sort of integrated with the kind of classical museum collections that doesn't science you. So I think that's going to get a really sort of major kind of challenge in that stuff. And it's happening. It's happening.

It's happening slowly. But I mean, like you talked about what would be the sort of great act to get. Well, we're at the point now where you can have these min-ion sequencing things, is it for USB things? Yeah.

The $1000 you can have a machine that pops into your laptop in sequence DNA. And you've got a PCR kit. But we're not that far away from citizen science DNA back-coding, which has been done. I mean, people set up the schools and various things.

So you mentioned people walking around back-coding stuff. That's going to be revolutionary. Would that be reliable to use? Could you be a trussing of citizen scientists to do this, do you know, Markle?

Good question. But I think the one other thing about sequencing is you can kind of tell what is from sequence. Providing there's not being a complete disaster with a machine, and you haven't got weird kinder sequences and things. You could give me DNA sequence and say it's something.

I could check it against databases and say no, it isn't. Whereas if you just told me you'd seen something, I kind of have to trust your judgment. So this could even be more reliable than I could see. But essentially, yeah.

Because you can actually go back. And so one sort of way to integrate this stuff is you have people have this sequencing and then you have people going into museums and using this kind of ancient DNA sequencing techniques to try and pull out, even if it's a fragmentary vets of DNA. So I guess one of the sort of dreams would be, you know, we've got this based on the museum. We've got a name for it.

You've got the sequence out here in the wild. I look at the same thing. There's a guy in Singapore who was basically back-coding the entire biota of Singapore. And he's doing it at the cost of, I think it's like a dollar of back-code.

And he's got this kind of amazing pipeline, mostly students. But they go out, they collect all these insects, for example, they sequence every insect in a collection. They put them into bags and these are sequences go together. These ones, they get a photograph, put a line, next, and away they go.

So you're providing machines themselves, don't make a mess of things. Or maybe I guess there's possibilities like, there are classic cases where you sequence what you think is the animal actually sequencing a plasti something like that. But if you've got a sequence, you can work that out. So I think some people are looking forward to just everything being sequenced because you can minimise it and sculpt it.

So taking a little bit of a turn here, it seems that their particular metrics by which we're measured is often a amount of grant funding that we receive, the number of high impact publications. But your work is clearly making major impacts in less traditional ways. So for example, you have a number of highly cited textbooks and bioinformatics programs. I'm wondering if you have advice for early career researchers like us in terms of how to strike that balance between the sort of traditional outputs that were expected to meet and some of these other really important impacts in other key areas.

So I'm probably the worst person to ask because you're exactly right, that's how we measure it and I don't play that game. I have very interesting conversations with them in every year. So what exactly do you do? So yeah, it's a challenge.

The other thing about advice, I think it's really hard to give advice because a lot of people get to it through luck or things just aligning at the right time rather than actually knowing how to do stuff. So some of the things I've done, it's been more or less kind of being there at the right time to do something. And I think one strategy is to make a tool or something that people find useful. So a few years ago there was a study of the most cited evolutionary biologists in Europe and I came top and I was gobsmacked.

Congratulations. But so I tried to figure out what is this going on because there are people doing fundraising things in evolutionary biology and it turns out it was a paper I write to display evolutionary trees. And in fact all the top five people were all people who built programs that people used to make trees and people cited them. So one strategy is you make a tool that people find useful, you make sure that you have something that people can cite.

I was lucky enough to write a paper. It's a two page paper, it's got a screenshot and a couple references, the contents, zilch, but it describes the software program. And so people cite it. And so I think that's one strategy.

You make something that people find useful, you give them away the site and you know where you go. So I guess that depends partly on can you make a tool or maybe if you describe a technique that people find useful, that's one way to do it. It is difficult and I guess part of the challenge is I think the culture is changing but very slowly. So the kind of people I follow in Twitter think most of these measures of economic output, like focusing on high impact journals, is all nonsense.

There are people who think of nature and science as being like the sun or the day to make up their kind of tablets. The real hardcore science has done elsewhere in a sense. There are people interested in measuring output for some sense of publishing data and getting citations and credit for data collection and publication. There's always kind of things happening.

There's things like metrics, measuring newspaper reports, tweets and so on. Which is great. And the problem is if you're then actually involved in any of the meetings about how we assist people that hasn't happened yet. So it's tricky.

So I think it's also bad to ask somebody like me because I think I have a little bit of a license in a way and that I have played the game. I have some highly cited papers that will be real ones. So I feel like it's going to be a little bit of a license to kind of do stuff that isn't like that. So if you're going to the rules of it you can end up in the meeting.

Well a little bit. I think it's a risky strategy because you know people look at this and on paper you're not playing the game properly. You're supposed to be doing other kind of things. So it's a risky thing to do as a young person.

But I guess that the challenge is is it going to be a case that by the time it becomes a shoot. They've got those sort of changed. And maybe they will. I think if you want to sort of get something time packed in a way.

You can do some brilliant research and get it published in ancient science while you're going to go for it. But the things that do seem to get highly cited are the sort of methodological things where you've done something that helps people solve a problem. People didn't respond to that. As I said, they're most highly cited papers and I've been listening to the politics in Europe.

I think all about people. That was a problem that people kind of had. I guess the one that called them more broadly would be things like the out packages. So some of the people have a huge impact on people who have written an out package.

So if you write an out package and you get it in a suitable form as well, that's great. People start to see you get citations and people see it as being a worthwhile kind of thing to do. It's still difficult. There's lots of people who develop software and feel the contributions aren't valued.

There's all those sorts of things going on. So either be brilliant in published ancient science or do something useful. This might be one way to do it. So on that note, it seems that you maintain an active dialogue with the scientific in-y through social media.

So you have over 3000 portfolios. How do you see the role of social media in shading your science not only to the public but also to other scientists. And have you found any drawbacks to actually engage in this way? It's a bit of a time sink but it's kind of fun.

It's my main source of news in terms of what's going on scientifically. And I guess what's kind of fun about it is you see all sorts of unexpected kind of things. So lately I've been interacting with a woman who is based in New Zealand. She's incredibly active in putting information about New Zealand plants and animals on Wikipedia and Wikipedia.

And she'd become, so that has been out of her New Zealand about the kind of amazing work she's been doing in the area. And I wouldn't have encountered her but for a few connections with various people in various kind of museums. And we've been interacting and collaborating on lots of information available on Wikipedia. So all sorts of things to come about just through making these kind of connections.

You learn about the most fascinating things as a guy who's got a website called Flyload Pick. And you may have seen the upper... Basically, if you ever see a paper and somebody's got a nice little silhouette or a plant for an animal, the chances are it came from this website. So he's built this website that if you want a picture of like a secret tree of cichlits or something and you want some pictures and you want a shark or a contrast, he's got these three openly available pictures.

He's crowdfunding to try and get money to release an exhibition. So it's on Twitter and I sort of sample across it and then you retweet it and you get other people's sort of engagements. It's still my major source of trying to find out what's going on and making sort of connections often I guess with people he wouldn't normally interact with. So I think it's invaluable.

I mean I don't care about Facebook or anything else. I think the thing I'm so slightly conscious of is I think people's experience of Twitter is probably quite different if you just join. There's some horrific stuff on Twitter. But I think it's the analogy I use with the students is if you go to a bar, lots of people, you can have lots of silly conversations.

But there'll be somebody similar in that bar who may have something interesting to say. You just got to find that person. I think Twitter's very like that. If you find the people who are doing interesting things and follow them and you follow their connections then and where you go.

But I find it invaluable. I mean there's other things that happen. Things like conferences. You can't go to a conference.

There's a Wiki later event happening in Berkeley next week. I'm teaching, Caco, rather on Twitter. And you can sort of follow along and see what's happening. So yes, it's a group of you.

Actually, ironically, I mean also it's a way that I find out what people are doing in the building. Because quite often there'll be a little thing that will pop up and I know that was going on. I think it's a great way to sort of just have information sort of flowing around and sort of discovering stuff. Well, this has been such a great conversation.

Thank you very much Professor Page and thank you for listening to this podcast and we hope to catch you next time. Thanks for listening to The Naturally Speaking Podcast. If you've got any ideas for future shows or want to get involved, please contact us. We'd love to hear from you.

Our email address is naturallyspeakingpodcast at gmail.com. Naturally Speaking is a podcast from the Institute of Bardress, the Animal Health, Paris and Medicine at the University of Glasgow. And more information you can find us on Facebook. You can find us on Twitter at theNSpodcast.

And you can download future episodes and find more information at our blog at nattressspeakingpodcast.webpress.com.