Molecular Human and the Multi-Omics Revolution: with Dr. Karsten Suhre and Dr. Anna Halama

Podcast published on 5/31/2023 • Show notes written by Vanja Maganjic & Rina Bogdanovic

53 minutes reading time
.
.
.

Episode summary

Are you ready to journey through the intricate layers of Omics, a rapidly developing field that's reshaping our understanding of biological systems? From genomics to proteomics and metabolomics, the multi-omics approach intertwines these layers, offering a rich, multidimensional view of human biology. Today's conversation unravels the fascinating complexity within our bodies, spotlighting the current advancements in multi-omics research and its transformative potential in disease diagnosis and treatment. You'll also discover why glycans, despite often being overlooked in multi-omics studies, are in fact promising biomarkers due to their role as intermediate phenotypes. In this conversation we are joined by two scientists from Weill Cornell Medicine in Qatar. Karsten Suhre, the Professor of Physiology and Biophysics and the Director of Bioinformatics Core and Dr. Anna Halama, who is an Assistant Professor in Research of Physiology and Biophysics. Tune in, as we unravel the enigmas of this complex and transformative field of research.

Conversation timestamps

We Discuss:

• Anna and Karsten's Collaborative Journey [02:27]
• Metabolism and Metabolomics [04:40]
• Linking the Dots: The Interconnected Nature of Omics Layers [07:44]
• Glycomics: The Underrated Player in Multi-Omics Research [09:26]
• Qatar Metabolomics Study on Diabetes and the Molecular Human [11:52]
• Unveiling a New Platform: An Introduction to the Comics Server [23:41]
• Intermediate Phenotypes [28:04]
• The advantages of a multi-omic approach to research [33:12]
• Overcoming the Hurdles: Current Challenges in the Multi-Omics Landscape [36:18]
• How Much Can Genomic Sequencing Inform Prevention and Diagnosis? [45:17]
• From Lab to Clinic: Karsten and Anna's Pursuit of Translating Research into Practice [50:21]

About the guests

Conversation highlights

"In genomics, you have just four letters, in proteomics, you have 20 letters, amino acids. And in metabolomics, the problem is like every molecule, I mean, not every molecule, you have acids and bases and all that kind of things, but it gets even more complicated. And that makes the measurement much more challenging. So there's not one tool to measure metabolites, there's more or less one tool to measure all proteins and proteomics. And there's certainly one tool to do all genomics. So, the more you go towards metabolomics the more complicated things become in terms of measurement."

 

"Right now metabolomics is very easy to think about. It's metabolites. Everybody knows about glucose, everybody knows about lipids and amino acids. This is something pretty natural to add to your studies. But not many people are aware of glycans. And also, if you think about technology, the aspects and the collaborative work. I mean, there are not many labs, which are specifically focused on glycomics, you need to have extra knowledge to do it right. So metabolomics right now, people can do on their own, there are some kits, which you can implement, you can do metabolic work, and you can also try to do something very easy for yourself. But with the glycomics, I feel it's more complicated."

 

"For the first time, we're presenting the Comics as a server on the third Human Glycome Project. It was the conference, which took place in spring, and people were really very positive about it. So I was also shown to some of the medical students of ours, and they're all like - wow, really, it's possible, can we look from that perspective? Because usually, it's all about the perspective, right? With this tool, we are kind of going out of regular paper and regular publication to another space, to another dimension where we are letting people interact with our results. And I think this is something very unique and cool."

 

"I mean, you always ask yourself as a researcher, why are we doing this? And I mean, on the one hand, it is just your natural curiosity. It's basic research, and we are geeky, but then when I talk to my mother and say, Okay, why are you doing this? I mean, then you really have to say - Okay, it's pretty tiny, what we really do, we are not, I always think of research as a pyramid. I mean, there are a lot of people at the base, doing a lot of stuff, they are very good researchers, but then they're never in the limelight. And they're all working together. And in the end, you have the few who are maybe just lucky to develop the RNA vaccine or things like that, that then really make an impact. But I tend to feel that they wouldn't have done that without the whole community around them and be a part of that. And that's what I think science is really about, and what makes it attractive to me."

 

Episode transcript

Rina’s Intro

Rina Bogdanovic [00:05] Hello, hello, and welcome back to GlycanHub - the podcast in which we explore health, disease, and longevity through the lens of glycobiology. My name is Rina, and I am your host. Omics is a rapidly evolving, multi-disciplinary field which allows us to gain a comprehensive understanding of the underlying structure, function and dynamics of a biological system. You might be familiar with some of the omics layers, such as genomics, proteomics, metabolomics and so on. A multi-omics approach takes it a step further. By integrating the data from the separate omics layers, we are now able to generate a multidimensional view of human biology that unveils the complex interplay of networks within our system. My guests today will introduce us to the current state of multi-omics research and its potential to revolutionise the way we understand, diagnose, and treat diseases. Listen in to find out why glycans are considered intermediate phenotypes and how this makes them promising biomarkers. As well as why glycans are often overlooked in multi-omics research. In today's episode, I am joined by two scientists from Weill Cornell Medicine in Qatar. Dr. Karsten Suhre, the Professor of Physiology and Biophysics and the Director of Bioinformatics Core and Dr. Anna Halama, who is an Assistant Professor in Research of Physiology and Biophysics. Karsten is a pioneer in multi-omics GWAS study. He shaped the field with his contribution to metabolomics and proteomics GWAS by highlighting their role in human health. His research focuses on complex diseases where integrative omics analysis could provide further insight into disease mechanisms and treatment strategies. Anna focuses predominantly on dysregulated cancer metabolism and its role in drug resistance. She is working towards strategies enabling the implementation of metabolomics and other omics technologies into clinical pipelines to support future diagnosis and optimise treatment modalities. A warm welcome to Anna Halama and Karsten Suhre.

Karsten Suhre [02:25] Yeah. Hello. 

Anna Halama [02:26] Hi. 

Anna and Karsten's Collaborative Journey

Rina Bogdanovic [02:27] So unlike the majority of our other guests, you two are not glycobiologists. So could you tell us what it is that you do? And how do you know each other?

Karsten Suhre [02:35] Maybe I should start on that. Because actually, I was her co-supervisor on her PhD, I think you said 15 years back.

Anna Halama [02:43] Yes. 

Karsten Suhre [02:45] Scary.

Anna Halama [02:45] We don't want to count time because it starts to be scary. No, it's fine if it’s one or two or three years. But then 15 years, you're getting… 

Karsten Suhre [02:53] We started together at the Helmholtz Center in Munich. And that's also where we both started, I think on metabolomics. I was more on the biophysics side, and Anna is more on the wet lab side. And that's how I think, then I moved to Qatar, and Anna did her PhD and I asked her to come, and maybe you can take it from there.

Anna Halama [03:13] Yeah, I think so. The story starts at Helmholtz Center as Karsten is saying, so it was an amazing experience, you know, we're always interacting together. And, in fact, Karsten kind of determined the next steps in my career, because I decided to follow him here and work in his lab. So the initial work was focused on metabolomics because this was the speciality I was trying during my PhD. And this was also the speciality of Karsten. So he was having this virtual metabolomics lab, and I was trying to add the wet lab component where we could, look at certain things, checking for the mechanisms and things like this. This is our story, this is how we started to work together. And we are working until now, it's like 15 years altogether when we counted this, which was kind of interesting.

Rina Bogdanovic [04:09] It sounds like you get along well.

Anna Halama [04:13] Yes, yes. Because, you know, usually it's something like, you're having many collaborators, but I think it's yes, it's very cool.

Karsten Suhre [04:25] We try to make it complementary. Because I have two left hands for whatever is coming to a pipette. And Anna is very good at working in the lab and with cell culture and samples and everything. So I think it's the complementarity that makes it. 

Metabolism and Metabolomics 

Rina Bogdanovic [04:40] Excellent. In today's conversation, we are going to focus on metabolomics but specifically also more on looking at the bigger picture and looking at a multi-omic approach to research. But starting with the metabolism. We have in our everyday life now started to use the word metabolism pretty frequently. Do you think people tend to misunderstand what is meant by that word, and how it works?

Anna Halama [05:04] I think people are using frequently metabolism in the way of - start to do exercises, you're gonna turn up your metabolism, change the diet, you're gonna reduce weight because you're changing something in your metabolism. And I think in that particular context, they are using the word metabolism very correctly. And they understand pretty well what metabolism is because everybody was having at some point in their life, biochemistry classes, when we're talking about, you know, starvation or making exercises and things like this. But we need to think about two different things. On one hand, we are having metabolism, when we are looking at the function of the body, on the other we are having metabolomics. And metabolomics is in fact, the study, which is enabling us to provide further insight into this metabolism. So it's a very powerful tool. And I think people are not really using metabolomics as a word, because not many people in the general population care about it. But metabolism as such, yes, they are using frequently, and in my view, they're using it correctly. So that's how I see it. I don't know how you see it's Karsten.

Karsten Suhre [06:13] Yeah, no, absolutely.

Rina Bogdanovic [06:15] That's a relief, that we actually are understanding in everyday life, what our metabolism is, but could you define more closely what metabolomics means?

Anna Halama [06:25] If you think about metabolomics, it's a pool of metabolites. So ideally, you are gonna be looking for all the small molecules in the sample. So not only on glucose, or lipids level, which you can do with simple biochemistry but looking at all at the same time. So this is how I see metabolomics. So studying the small molecule composition, in the given sample, and whatever the sample is, it can be a human sample, it can be dates, I mean, the story on dates, or it can be everything, whatever you can think of. 

Karsten Suhre [07:04] It's just like genomics, proteomics, and metabolomics, you just do metabolism as a whole and of course, it gets more and more complicated. In genomics, you have just four letters, in proteomics, you have 20 letters, amino acids. And in metabolomics, the problem is like every molecule, I mean, not every molecule, you have acids and bases and all that kind of things, but it gets even more complicated. And that makes the measurement much more challenging. So there's not one tool to measure metabolites, there's more or less one tool to measure all proteins and proteomics. And there's certainly one tool to do all genomics. So, the more you go towards metabolomics the more complicated things become in terms of measurement.

Linking the Dots: The Interconnected Nature of Omics Layers

Rina Bogdanovic [07:44] Could you perhaps give a small overview of how the different omic layers interact or relate to each other?

Anna Halama [07:49] We are having obviously, all the omics, which Karsten was mentioning. So in fact, all there, they are linked together, right? Because you are having from genomes, we are going to transcriptomes because of the transcription. The transcripts are later on translated to proteins or enzymes, transporters, you name it, and then metabolites are products of the substrate for the function of those proteins. So this is the way how all of those are interlinked. Then there is another factor, which is that not all enzymes are gonna be working in the same way. And they sometimes change, for example, with glycans. So you are having glycosylation, and then this glycosylation might change the function of the enzyme. Also, if you think from the perspective of genomics, you can turn on or turn off the genes using methylation. So it will be another way, adding the complexity to all the omics layers, which are together. I would say that everything connects with everything else, which makes it fascinating, in one way or the other. If you think about complex diseases, it's pretty difficult to interpret later on and work with those data. So this is how I would describe it. 

Karsten Suhre [09:09] Some of our early collaborators from the company Biocrates had, at some point, this image thing to say - in the end, the more phenotype you get, the more it's like imaging, right? In the end, every single omics phenotype is a pixel in a picture and the more you get, the sharper the picture becomes.

Glycomics: The Underrated Player in Multi-Omics Research

Rina Bogdanovic [09:26] You have mentioned glycomics as being one of the omic layers. Why do you think it is often overlooked in studies that are taking a multi-omic approach?

Anna Halama [09:36] It's not that popular. I mean, think about the collaborators, right and think about possibilities. Right now metabolomics is very easy to think about. It's metabolites. Everybody knows about glucose, everybody knows about lipids and amino acids. This is something pretty natural to add to your studies. But not many people are aware of glycans. And also, if you think about technology, the aspects and the collaborative work. I mean, there are not many labs, which are specifically focused on glycomics, you need to have extra knowledge to do it right. So metabolomics right now, people can do on their own, there are some kits, which you can implement, you can do metabolic work, and you can also try to do something very easy for yourself. But with the glycomics, I feel it's more complicated. And I think that this is one of the reasons for the complexity of the story.

Karsten Suhre [10:39] Right, and also metabolism, you have like 150 years of biochemistry already going on. Even if people know about glycans, I mean, Gordan Lauc is I think very outspoken about the Human Glycome Project, and people really underestimate the role of glycans. I totally agree with that. But the limitation is really - how do you measure this in a straightforward way? We got some data actually from Gordan. But I mean, it's still very limited. You just have glycosylation points on one protein of IgG, for instance, or you can get all the N-glycans, but that would be just 30 N-glycans on any protein. So it's very hard to measure everything, especially if you go for larger studies, population studies, where you have clinical outcomes. So you may look at this in much detail if you do a study in cell culture, or if you look at a specific process, but to do it at the population level, where it would work like what we do with proteins and metabolites and genes. I think the biggest limitation at the moment is the measurement to really get this data to really show what the impact is. And I think Gordan has done a great job of showing examples. But I think there's much more to come.

Qatar Metabolomics Study on Diabetes and the Molecular Human 

Rina Bogdanovic [11:52] That is a very common answer when I asked a similar question why glycans aren’t more researched, that it’s the technological limitations. Now segwaying, to the Molecular Human. Could you tell me about the Qatar Metabolomics Study on Diabetes? What were you initially aiming to do with this study? And how has it grown into the Molecular Human?

Karsten Suhre [12:12] I came to work in Qatar almost twelve years back. And at that time, I came from Munich. So we have been doing metabolomics on other kinds of things there. So the idea would be, we would like to do it here again. But then the challenge was where are the samples? There was no Qatar Biobank at the time and no KORA study and no Framingham study for Qatar. So we started what we call the Qatar Metabolomic Study on Diabetes, initially, with the aim to send it to just total metabolome, because that's the big study we did at the time. And we said, we collect urine, saliva, plasma, which was a little bit I mean, what's been inspired at the time, Qatar Biobank was just building up and my colleagues from Munich were coming here to consult with them, and also people from Imperial College who are here to, to basically build this up. And we said, maybe do a little kind of a pilot for Biobank study. And so we said, we would do a diabetes study, because that was one of the priorities in Qatar at the time, because there's a high diabetes rate, as you may know. So we said, we go and collect 50 samples cases, 50 samples controlled because we just had done that in KORA before in Munich. And we were just over our luck that in the end, we didn't turn out with 50-50, but with basically almost 400 samples, and we also, it was really serendipitous in a way because we wanted to do that with the diabetes department. But there was no space, they were so busy. So in the end, we turned out at the dermatology department, because there are many diabetic people with skin disorders. But you also have a lot of people who have minor things that you could actually use as controls. And we really collected almost 400 people. Totally mixed, so there was really no batch between cases and controls collected in different places. And I think that's very important. And we were lucky with that. And we created I don't know how many 10s of aliquots per individual. Initially, we just wanted to do metabolomics. And then really over the last 10 years, every time a new technique came up, we said oh - let's just try it on this pilot. So we have a great Genomics Core here. They ran genotyping arrays and methylation arrays, and we did RNA sequences, and whole RNA sequencing on them. And then at some point, I mean, I was in contact with Gordan Lauc, who basically was already doing the measurements in KORA. And we just said - oh, why not send them to Gordan as well, and then maybe we let Anna continue with that because we have a lot more data in there. And Anna is just presently trying to integrate it into one. So maybe we’ll go on with that one.

Anna Halama [14:55] Yeah, stories like this. And also it will be important to mention that at that time - the information on diabetes and the diabetic project was not there. There were not so many studies at that time when you started to work on diabetes metabolomics and stuff, because it was 2013 when you got the data. 

Karsten Suhre [15:13] Yeah. 

Anna Halama [15:14] So there was also not so much information about diabetes in terms of metabolomics. At the time was also the new metabolomic platform where the study was. So all of this was touching upon the technical development as well. But I think once the Olink came to the market, and we tested it as new technology also, just from curiosity, we started to look at the data in more like the way where we are trying to compare two different technologies, where we are looking at different data sets. Of course, the methylation with diabetes and stuff, all of those publications were coming down the line, so there were many, like PWAS [Proteome-Wide Association Study]  papers. But on one point, we look at this one and said - Well, I mean, it's the very first study, which is looking at so many comics at the same time. We have 18 platforms now.

Karsten Suhre [16:18] Because I think what's maybe also important to know is I mean, here Weill Cornell in Qatar is very new, compared to the hundreds of years of history of Cornell is very new. I mean, I think that the research started in 2010. And we have something which is very important, our different cores - Genomics Core, Proteomics and Metabolomics Core, Bioinformatics. And we use this study, basically, as a trial bed for all the different tools, we had the SomaLogic platform, and at some point, we brought the Metabolon platform to Qatar, Biocrates platform we have briefly here. So there was like this combination between logistical and testing, then bringing it into studies. And now what Anna is doing, maybe you can continue on that one is now how can we integrate this? Because of the question in the aim, the aim was not to integrate multi-omics at the start. That's what we are now doing.

Anna Halama [17:10] Yes, exactly. Because for me, it's like, can we do this because, I'm a biologist. So I'm looking from a different perspective of the data. And I don't like to look at the Excel sheets. You're having p values, okay, you understand them, but what do they really mean? So I was always coming and kind of bugging Karsten with questions - Can we do this? Can we connect? and then he was coming up with the idea of this - mutual best hits, to connect the data in the first thing, and this is the way it started. So from the mutual best hits, we were starting to create the interaction matrix between first of all the molecules, but also different omics platforms. And, this is how we thought - hold on, I mean, usually you're having the single layer, but here we are having like 18 different platforms. So you're kind of describing the human body with this case-control study, on the very much molecular level where you're super specific. And this is the way how we thought about Molecular Human so it's coming a little bit from Leonardo da Vinci you know, the print, where you are having the Vitruvian Man. Where you look at that the ideal way of presenting human beings. So we were thinking that with all of those platforms, we are able to create the molecular representation of the human being.

Karsten Suhre [18:38] And what's maybe also important is I mean, I'm a mathematician, so there's also like, kind of algorithmic stuff in there. Together with my colleagues, actually also from Munich at the time, Jan Krumsiek. He's now at Cornell in New York. So he was the guy who basically brought Gaussian graphical models or what is called partial correlations to metabolomics. And then later, he actually used it also on glycomics data. So in this Molecular Human, we have like a way of integrating data inside of the platform using these Gaussian graphical models, and then what Anna mentioned, the mutual best hits across platforms. So the idea of the mutual best hits briefly is an idea that comes from basic orthology in bacteria, where you say if I want to compare the function of two genes between two bacterial species, I want the best blast hit in that case, from one genome to the other is the reverse best hit from the other genome to the first and with the same idea if I now do mutual best hit between a glycan and a metabolite, I would only link those that are really mutual best hits and say these have something in common. So if they are from a similar platform, they could actually be the same molecule if I compare SomaLogic to Olink or Metabolon to Biocrates, but they could also be something, they are basically carrying the same kind of information or that they may be something different. And in this sense, the Molecular Human and then the web server that we call the Comics server behind that, (and maybe we should talk about that in a moment), is basically a tool to actually look at this integrated data because I don't think you can integrate data automatically, you need to bring it together, put it into a visualisation tool, and then you have to use that to interpret it.

Anna Halama [20:23] Because interpretation is the biggest problem. If you're having so much data, what are you gonna do with that? So what will be your next step? How are they interacting? These are all the questions which we were asking while working on this project. Because I mean, it's not something you're doing over the nights, we were having 1000 discussions and back and forth, and then, you know, understanding of how those molecules are interacting, but what these interactions really mean. So our initial goal was to see if we can replicate whatever was previously shown in the literature, and then go further, to what is new. But what is key to this entire project is connecting omics, which is our Comics server, right? We are seeing it as a window to the Molecular Human. So there, you can have a look at the processes of the interaction between the gene and the molecule and methylation, for example, it's also linked with the GWAS catalogue. So it's a pretty robust tool, which is available for everyone. So whoever is interested in some protein, and it's usually looking in the silos, not considering the rest of the processes, which are ongoing, can right now go and check. And I would strongly encourage people to do so because for me, being a biologist, it's great fun. You're having your protein or metabolite of interest, and you can understand how it's connected in the entire network. And I see it as a great plus, for the entire scientific community. That's why I mean, we are sharing, not keeping it for us only. Because we would like to, you know, give the people the opportunity to experience what we were experiencing while working on this project.

Rina Bogdanovic [22:09] Just to clarify, you said you were using 18 platforms, and how many omics were you looking at? 

Anna Halama [22:15] I think it's eight. So if we're thinking genomics, then it's methylation - so epigenetics, then we have transcriptomics, micro RNA, then we had proteomics, glycoproteomics, glycomics, metabolomics, and lipidomics. 

Karsten Suhre [22:32] Yeah, so 9.

Rina Bogdanovic [22:35] So some of the platforms were used for the same omics?

Anna Halama [22:38] Yes, correct.

Karsten Suhre [22:40] The thing is like, you cannot really draw a line, for instance, what do you call Biocrates’ data? They have a lot of lipids on there, but also Metabolon or Nightingale, do you call them the one or the other? So I think there are 18 different platforms. And then also we have data in urine for some of them and in saliva as well. So in the end, there are 18 different measurements, and we have two more that are presently in the making. So we have also mass spec proteomics from the Seer company coming and from another company antibodies against bacteria. But that's still not published yet.

Anna Halama [23:17] And it's not on the server yet. But for instance, if you ask about the number of platforms, which were used per omics for glycomics, we have three platforms. So, of course, we were checking how good they are correlating and stuff. And the results, which we achieve were very good. So we were very happy with the correlations which we saw.

Unveiling a New Platform: An Introduction to the Comics Server

Rina Bogdanovic [23:41] I was curious, what was the response of the scientific community or collaborators when you released Comics? 

Karsten Suhre [23:48] Well, it's pretty fresh. I mean, officially that it really has the name Comics, and the Molecular Human that's like maybe a year that Anna is working on the paper giving it names and things like that. The server, I started it during COVID, actually, when I was just blocked, and just wanted to do some coding. Before that, I mean, we already shared the data and many studies, so it was used in different GWAS. And we also have a paper together with Gordan Lauc where we compare or combine SomaLogic with the N-glycans, for instance. So this kind of integration, so it's not something direct, and there are many colleagues from mostly from KORA that very often come to us and say - Hey, you have data on this, can you replicate also with the Twins UK, we're working on replicating things. So the QMDi study is a little bit too small to be a standalone thing for big GWAS or something like that. But if you want to do replication, the nice thing is also it's not Caucasian, there are Indian there are Filipinos in there. So whenever you see something in QMDi, you know, it's really robust. And that's also something which I find interesting because people would initially criticise it like, say, hey, look, these guys are not all super fasted, and they are three different ethnicities and whatever. But the thing is, if you replicate in that cohort, and it replicates there, it means that the signal you're talking about is a signal you can carry forward in other studies. And on that side, it's positive.

Anna Halama [25:22] I think I can also add to the Comics as a server because, for the first time, we're presenting the Comics as a server on the third Human Glycome Project. It was the conference, which took place in spring, and people were really very positive about it. So I was also shown to some of the medical students of ours, and they're all like - wow, really, it's possible, can we look from that perspective? Because usually, it's all about the perspective, right? With this tool, we are kind of going out of regular paper and regular publication to another space, to another dimension where we are letting people interact with our results. And I think this is something very unique and cool. And this is the way how people which I was talking about it with, were also saying this is kind of a very cool way of showing the data and putting it in a resource, this kind of data. Because it's not a big deal to bring a table. But here, you're having much more than that. And this is why I think it was not simple coding because Karsten was saying - I wanted to code. No, it was not like this, I mean, it's something more and something I would just really want to go to interact with and see how good it is.

Karsten Suhre [26:45] And I also, still see it a little bit as a pilot for bigger studies, especially when I now see UK Biobank with 300,000 samples from Nightingale. Of course, we cannot compete with that, but in QMDi, in Comics, you can see how the different platforms are complementary to each other what correlates, and you have it and I think you can also use it to actually evaluate, what would you be able to get if you go to a 10 times bigger study like KORA or 1000 times bigger study as the UK Biobank, and I think it could be very motivating as a driver. And I think I may even shared a very long time ago, some of the Nightingale data with them to see how Nightingale compares in terms of coverage to Biocrates, Metabolon, things like that. So you know, the complementarity. And also, when you discuss, what do you want to do? It's not like, oh, there's just one metabolomics. I mean, there are three, four or five platforms, and they are all very different. And then, in the Comics server, you can actually see this complementarity and also the similarities for the mutual best hits, what is actually overlapping between the platforms, and what each platform has on its own.

Intermediate Phenotypes

Rina Bogdanovic [28:04] So in the Molecular Humans study, you describe glycans as intermediate phenotypes that provide insight into both the genetic background as well as the potential mechanistic link to clinical outcome. Could you explain what is meant by the intermediate phenotype? And how it might be advantageous in a clinical setting?

Karsten Suhre [28:25] I think the concept is to give credit to Florian Kronenberg, who is a colleague of mine who was on our first GWAS paper with metabolomics in I think 2008. So that study really saw this idea of intermediate phenotype to say, I mean, at that time you did GWAS between disease endpoints, so say - Oh, this gene is associated, or genetic variants in this gene is associated with the risk of diabetes, but you didn't really know why. And the idea of the intermediate phenotype is to say, on the one hand, you have a correlation with the metabolite and the disease, which is strong. And then you have also a strong association between the genetic variants and the metabolite. So in that sense, it's an intermediate phenotype. And also, there may be many pathways that all go towards diabetes, and depending on which metabolite, and I mean, now, it's a metabolite that could be a glycan could be a protein as intermediate phenotype, which ones are the ones that are driving this or that pathway? So the idea of intermediate phenotype is really like I mentioned before the imaging the fine mapping of these pathways, so if they’ve been in association, and it doesn't have to be necessarily genetics, it could be lifestyle factors, smoking, things like that. So whatever you have in deep molecular phenotyping, is what I would see as an intermediate phenotype. And that is really working out very nicely. And that's the idea of why we do all these multi-omics things.

Anna Halama [29:54] I think from the perspective of our publication, we can say that we saw this for instance, because of the relation in the context of cancer and different glycan structures, obviously, this is something which we need to still follow up on and look to further validate it and to further look for the mechanism. But this is something which can be observed, the interaction between the protein, the methylation, the glycomics profile, and the link with, for instance, inflammatory mediators and cancer. So, you'll see that there is interaction, you'll see that there is a link, but then the next question will be, what does it really mean? And what are the mechanisms? And what is the driver of these changes and how it's impacting, let's say cancer, if we're looking from the perspective of cancer, and obviously, to put it into clinics, it's a long way. So. Yeah, I mean, it's a hint. 

Karsten Suhre [31:01] And I mean, just as you said, put into the clinic, I think the challenge, also, sometimes the frustration is the translation. I mean, how can we really make sure that what you do here has a real impact and is not just some kind of academic paper that nobody reads at the end of the day? That's really nowadays the big challenge of how to translate that, and I don't have a very concrete answer. I think there's not one answer to that. 

Anna Halama [31:27] I think the entire process is ongoing, there is a huge gap, the one which I see because right now, I'm trying to be even more involved in the clinical study working on cancer with the physicians. So there is a gap between the researchers and the environment of the clinic. So, if we are thinking about translation, first we need to overcome this gap right? How you will want to be doing the samples, I mean, if you like to do clinical research on multi-omics, you need to have an entire infrastructure you cannot come and do research core. You need to have the infrastructure of machines, you need to have skilled personnel, and then the readout, right. Because I mean, we are showing Molecular Human we are having so many omics layers. And I mean, the interpretation of those data was a challenge for us initially. But if you are a medical doctor, and you are seeing this massive amount of data, I mean, how useful it can be for you? So I think before we are even moving towards the clinic, it will be a pretty long way. And it needs to be also thoughtful about everything we're going to be doing as a first, second, and third step to make it possible. Because I mean, right now, I see that it can be possible eventually, but not next year or not in two years.

Rina Bogdanovic [32:50] Yeah, absolutely. That sounds like an ideal scenario, we can have all of these tests, have a big picture and the doctor can understand exactly what's happening in your body through this elaborate system. But obviously, it will take a long time to enable this.

Karsten Suhre [33:05] You throw in the buzzword AI, and everything is solved, and you don't have to do anything anymore. But that's ironic, right.

The advantages of a multi-omic approach to research

Rina Bogdanovic [33:12] But let me just go backwards for a second, just to perhaps clarify what is the advantage of having a multi-omics approach when compared to perhaps looking at single molecules when we're trying to understand complex diseases.

Karsten Suhre [33:28] So I think the idea of having multiple omics I mean, you want to look at all the steps that play a role. And as you mentioned before, glycans are in the moment, we know there could be a bus coming, but we don't see them because we don't measure them.

Anna Halama [33:42] Also, pieces of information. When you're mentioning those pixels, if we think about a piece of information, like one molecule will be one pixel, but I mean, it's not bringing you to the entire picture. How many pixels, do you need to see this picture? And, okay, we were saying that all omics are connected. And it's fascinating that the body is functioning with all this complexity. But one, if one system is starting to get broken, you cannot really look at a single molecule. I mean, even when you look at diabetes, for instance, if you were gonna look at the glucose, it's not enough. 

Karsten Suhre [34:25] And I think it's a trade of at the end, of the resources you want to spend and what you can see and I think for that, I think our study could be very helpful because it's very deep, but you can get a feeling for how deep do you really need to go for your specific understanding and then focus on things that may be relevant for the one or the other phenotype you're looking at.

Rina Bogdanovic [34:50] Now you have built this molecular network. Do you think it can potentially give you hints towards molecular interactions which are important in certain diseases but haven't yet been explored in that context?

Anna Halama [35:03] To some extent yes. And this was something which we're highlighting in the manuscript. So again, with the glycans and the interaction with the immune system, or introduction to the immune system, and cancer and in the context of cardiovascular diseases, also in the immune system. So those kinds of things we were seeing, some of them we’re replicating. That's why we know that the signal, which we see is somehow reduced, and not only novel but also true. But again, it will be worth it to make some verification of these findings.

Karsten Suhre [35:41] I think the weakness of our studies, of course, I mean, it's a case-control study of diabetes, but we don't have follow-ups like this big UK Biobank and things like that. So I think what we really need in the future is phenotype incident disease, this kind of thing. And that's why I'm very much looking, to see if we can use this to motivate studies like UK Biobank to have even more omics data on that. And then from there on to say, okay, which kind of correlations do I see at time point zero? And which of these are actually correlated to an incident heart attack or something like that? Maybe a few years on.

Overcoming the Hurdles: Current Challenges in the Multi-Omics Landscape 

Rina Bogdanovic [36:18] You have spoken about the challenges of translation of this work, but overall, what do you think are some of the main challenges currently in the area of multi-omics?

Anna Halama [36:28] I would say it all depends on the perspective. Because when I look at the challenges that will be probably different than when Karsten is looking. So I can tell you like, I see that there is a need for more omics data, right? And also willingness from the scientist. But the preparation for such a study, it's maybe not still there. I mean, you know, the very important thing is to collect the data in the highest possible quality, so that the bioinformaticians don't have to struggle. But I see that the first struggle with multi-omics is starting initially with the sample collection and experimental design. So this is something which should be kind of clarified once and forever, with the processes in place that everybody is obliged to follow in order to produce good quality data. So the first challenge is in the beginning, then the entire bioinformatics is happening when Karsten can comment on this one, and then we are having these huge datasets. And it's very difficult to interpret, I still see an interpretation of the data, when you have so many different factors you need to look at, it's kind of challenging with the Comics, we're able to overcome some of the problems. But there are many stories, which we could talk about. When we were starting initially, the manuscript was huge, because wherever we were starting to look at there was coming very interesting story for us, but not necessarily for the reader, you know, so structuring your thoughts and making the stories another thing when you are facing this amount of the data, and also how to connect them is another thing. But there are also other things which are happening in between.

Karsten Suhre [38:23] Yeah, I mean, there's one level where it actually absolutely happens. And that's genome-wide association studies. So if you take UK Biobank, the pharma companies are putting millions and even 10s of millions of dollars into getting proteomics data and maybe other data sets in the future, a whole genome sequencing exome sequencing in UK Biobank and the pharma companies are paying for that. And they're not doing this for any altruistic reasons, but because they find genetic evidence for drug targets. And that's what it's all about. And I think that's a huge translation that you have there. It's not very much publicised, because what in the end, the pharma company takes out of this and what decision-making they do on their drug development, that's not something that's in the public domain. But I think that's a very realistic translation and measured in the kind of money pharma companies are putting in there. So I would say this translation, their translation to the clinic, I mean, there's another misconception maybe just say, I don't think that you would ever have multi-omics measurements on a single patient. It's more about what you would like to find in studies and the biomarkers that are relevant. And once you have them, then you measure the relevant biomarkers in the clinics. Although in some cases, I mean, genomics if you do whole genomes on cancer patients, and I can imagine that maybe at some point, you would do a whole metabolome on cancer on tumour samples to find out what their metabolic weaknesses are. And I think maybe Anna, you had this one project, maybe you can comment on all that kind of information.

Anna Halama [40:01] Yes. This is something which we are trying to do right now. Because I mean, if you think about tumours as metabolic disease, and you would like to understand how you can tweak tumour metabolism in order to support the treatment and things like this, you will have to understand what metabolic processes are upregulated or downregulated in this tumour to find the way to find the hint. So, to translate it to the clinic, you need to have a real patient, you need to have a real tumour, or you need to have a real way of extracting the samples from a really very small amount of the tissue. And this is something which we are doing right now. So, if we will be successful in this approach, we're going to have, let's say pipeline for multi-omic characterization of the tumours, right, and also characterise them metabolically to attack the specific metabolic dysregulation, which you can see in cancer. So this will be the way of adding an extra layer and why we're looking on this one specifically, because there are already companies which are in fact, attacking tumour metabolism in patients in clinical trials, which they are going for glutaminase metabolism or fatty acid synthase metabolism, which is, again, very relevant. The question is of course - are there some feedback mechanisms from the tumour, and do we know how we can look at them? And this is something where we were studying together with Karsten, on the metabolic engineering approach. So this was, in fact, our very first day together on cancer when we were drafting the grant, which was awarded by Qatar Foundation, and we were trying to look at how we can tweak the metabolism of tumours to find the feedback mechanism and to attack both pathways simultaneously. And we have a very good example with glutaminase, which is also known that some cancers are addicted to glutamine. So our question at that time was, but what's happening then, is the cancer cell dying or not? And we find out that not the cancer cell is not dying. And in fact, it's activating other metabolic programmes, which because we're using the time metabolomics, we were able to detect and rationally target the two pathways at the same time. So I mean, this is the way how we are starting and how it can be leading further on to some translation.

Karsten Suhre [42:42] I mean, that will be the thing where you would maybe go with multi-omics into the clinic, if you really have like, especially in cancer, I mean, each cancer has its own set of mutations. I think, for other complex disorders that have one cause you don't need 1000 parameters. But if you go, for cancer, if you really want to understand the background, what's happening in that cancer to rationalise, what are the escape mechanisms of cancer to the first level to the second level treatment. So I think especially for recurring and recalcitrant cancers there, for instance, and multiple omics approaches are definitely interesting. And I think the other thing would be similarly monogenic disorders, which is also something we have a lot of here in Qatar, where you would have really very rare diseases where there's actually not much knowledge there. So even if you would find the variant, to actually understand what's physiologically happening in the patient, that will be very individualised. If you wanted to have all the omics data and then tell the clinicians - look, this is basically what's happening there, then you would need of course, a physician who has enough experience to know how they can tweak the one or the other pathway with an existing drug in these rare disorders. I think that these are the two things where really the omics could be directly translated to the clinic. But still, the challenge is the measurement because you need a measurement that's really quantitative reproducible, not something with batch effects that you can record.

Anna Halama [44:14] Something which is easy, you can put in a clinical setting. You need to find somebody who's going to operate this machine, because, you know, in proteomics and metabolomics in glycomics lab, they're experts who are troubleshooting the machines which are developing new ways of optimising processes, but in the clinic, you have to have something which is stable. If you think about all other processes, which are ongoing in clinic and the biochemistry measurements. So this is something which is established for years. There is, I mean no deviation unless there will be some problem right happening, but usually, it's a pretty stable measurement. In the case of metabolic omics or proteomics, I mean, it's still not very much, let's say quantitative or not batch dependent, and all of the other factors which have to be taken under consideration. So processes which are easier need to be developed to be translated.

How Much Can Genomic Sequencing Inform Prevention and Diagnosis? 

Rina Bogdanovic [45:17] Absolutely. And I'm just wondering when thinking about currently, when we mentioned different layers of omics, some of them are more available to us as consumers already, such as sequencing a genome. So I'm just curious, could you explain how useful looking at an individual genome in isolation without other omic layers would be in terms of prevention and disease risk?

Karsten Suhre [45:43] I would say it's already extremely successful. But there's definitely more to come. I think, especially if you identify things that you know what they are, then the genome is sufficient, right? If you identify a p53 mutation, you don't have to question it. But if you find variants, we don't really know whether they're really functional, whether it's really the variant that plays the role, then you need extra phenotype to validate that they are actually the ones that do it. And that's, especially in these more complex cases, like cancers and monogenic disorders.

Anna Halama [46:19] Also, I think, another step to add to this is, if you are looking for treatment. Then you need to, of course, understand how this gene is changing the processes, which might be optimised. So if you have only genetic information, it's very difficult to see the processes, which are changing, I'm not sure. So then you need to have all the other factors to think - Okay, I have this problem. It's causing this particular phenotype, what are the intermediate phenotypes which I need to, or can I tweak with the drugs, what's changing in the diet to make it, let's say, more feasible to live with this particular genetic problem?

Karsten Suhre [47:07] And of course, the other thing is, I mean, your genome is stable, right? Except for mutations in cancer, you just have one measure for your lifetime. While the metabolomics and the proteins, they are very much changing on different timescales, it could be fasting, non-fasting on a daily scale, it could be over a few weeks when you do exercise, and even longer. So I think what makes it also challenging is that genomics data is easy to deal with, because it's four letters, and they basically stay forever. The metabolome, in addition to all the exponential things that, maybe in the same measurements could give you different results, depending on how you tune your machine. But even if you would have the perfect machine, they still depend on so many extra factors. So the analysis is much more - Yeah, I mean, you need to put them into context. And it's the one hand, it's, of course, nice if you go for even for GlycanAge, for instance, the parameter you derive at Genos from that, I mean, that's just one parameter. But that parameter changes with activity with all kinds of things. And that's true for every single parameter we measure. And getting a handle on that it's also a cost factor. There are these studies where you have like, hundreds of measurements on just one person. That's the other extreme. And, and that's also extremely interesting. But getting this kind of detail from everyone is probably also a  impossible to be done. And I think what needs to be developed and that's really the challenge is some kind of tools for instance, there's this glucose monitor that you can put on your arm, if you would have something that would read out maybe 20 or 50 parameters, and that could be a combination of glycans and metabolites and proteins. And we choose the right one, I mean, that, for instance, would be something that would be like the Star Trek thing. You point to a person and you get a comprehensive understanding of what's happening. I think that's certainly the dream where we might go one day. And I think that's the path we're on. And, I mean, we should just be realistic with ourselves where we stand on the path and it's moving forward. I think we're making a lot of progress, especially these last years. It's a pretty amazing time to be in the field.

Anna Halama [49:30] Yes, I mean, that's for sure. 

Karsten Suhre [49:33] And also to move away from metabolomics and we started with metabolomics. But now having all the other layers and getting always exposed to new challenges, and also understanding what glycan data really is, was for me also a discovery because initially, I thought I would get something like omics metabolomics data from Gordan, but then I realised - Oh, there's just one spot on one protein where I have already like 40-50 parameters, and it's just one. And the other day he was offering me two more proteins. So I hope we have the opportunity to maybe generate that data as well. And then three proteins, but I think there are at least 100, or maybe even 1000 proteins in the bloodstream that are glycoproteins, that would be interesting to look at with these kinds of tools.

From Lab to Clinic: Karsten and Anna's Pursuit of Translating Research into Practice 

Rina Bogdanovic [50:21] Absolutely. I think a nice way of concluding this conversation would be to maybe talk about your efforts currently in the work of translating your research into the clinical application.

Karsten Suhre [50:33] Should we say they're classified? 

Anna Halama [50:37] Well, no don't. But I think the study that Karsten was doing together with our colleague Shaza B. Zaghlool on the subtypes of diabetes, I think this is something, which is pretty clinically very much relevant.

Karsten Suhre [50:51] Yeah. Exactly how things fit together because it's not something we did. I mean, there is a very highly cited paper that came out in New England Journal, I think, like two or three years back, or maybe five. And it's very highly cited, and clinicians are jumping on that. And they define basically four subtypes of type two diabetes. And what we did is we first of all tried - does the same thing apply to the Qatar population, and it does. There are other studies showing it in other populations, that's really something global to humanity, and not just for Caucasians. And then we try to do metabolomics, and proteomics on them. And you can find there are really very specific metabolites and proteins that basically go with these subtypes. And you can understand how these subtypes work. And now we're just trying to figure out, I mean, there are people already doing clinical trials to say if you're a certain subtype is a certain treatment for you better than another one. And that's, of course, I think we realise how much time all this takes. And once you come up with an idea, if you really want to bring it to the clinic, someone says - oh, clinical trial. And then it's like - years. Even if you get the money. I mean, it takes us years to get the funding and it takes years to do the study. And then you end up with something that, in the end, you would say was evident from the beginning. But before you can really bring it to the clinic, you still have to go through that. And I mean, we were. I'm still amazed how that worked with a COVID vaccination that we really got in one year. Would be nice if we could accelerate the translation of other things as well. But of course, the urgency is not the same as for COVID.

Anna Halama [52:33] Yeah, from my perspective, I'm very much focused on cancer. But this is something which we were just discussing, you know, with the tumour and multi-omics from the more how those things are laying how this is linked to the phenotype of the patient. I mean, all of those questions I'm asking in the study, which I'm currently working on. So this is something which is a work in progress.

Karsten Suhre [53:04] I mean, you always ask yourself as a researcher, why are we doing this? And I mean, on the one hand, it is just your natural curiosity. It's basic research, and we are geeky, but then when I talk to my mother and say, Okay, why are you doing this? I mean, then you really have to say - Okay, it's pretty tiny, what we really do, we are not, I always think of research as a pyramid. I mean, there are a lot of people at the base, doing a lot of stuff, they are very good researchers, but then they're never in the limelight. And they're all working together. And in the end, you have the few who are maybe just lucky to develop the RNA vaccine or things like that, that then really make an impact. But I tend to feel that they wouldn't have done that without the whole community around them, and be a part of that. And that's what I think science is really about, and what makes it attractive to me.

Anna Halama [53:57] Yeah, that's how it is. Because at the end of the day, my motivation is always to think how we can help,  but I am not as naive as in the past, because many years ago, when I did my master, I thought that with the method which I was gonna be doing, I will be solving the cancer problem, because we were working at the time on non-virial gene delivery. And this was something well, for sure it was gonna work. I mean, not everything is always like, I mean, it was working. But you know, that you need to have so many different things on the way and, of course, with time you'd like to contribute. This is the most important and, of course, making some kind of progress. And I hope we'll be able to do it with the things which we are doing.

Rina Bogdanovic [54:43] I will certainly continue to follow your work and find out more about these new studies, which are somewhat classified still. 

Anna Halama [54:48] Thank you. 

Karsten Suhre [54:51] All right. Thank you. 

Rina’s Outro

Rina Bogdanovic [54:52] Now speaking to our listeners, I hope this conversation gave you an insight into the tremendous potential that the multi-omics approach holds for the future of personalised medicine and human health. If you would like to access more information about this conversation, as well as Anna and Karsten’s previous work, follow the link in the description to the show notes for this episode. Equally, if you want to find out more about GlycanAge, head on to glycanage.com where you can access a whole list of our scientific publications, blog posts, testimonials and of course this is where you can order your GlycanAge test kit. Watch out for our next episode where I will be joined by Ronald Schnaar, a Professor of Neuroscience at Johns Hopkins University whose research focuses on neuro glycobiology. We will explore the role of glycans in the healthy structure and function of the brain as well as what glycan research can tell us about neurodegenerative diseases such as Alzheimer’s Disease.  Please don't forget to leave ratings and reviews for this episode and engage with us on social media. Thank you for listening and have a great day.

 

Disclaimer

Please be advised that this show is for information only and should not be considered as a replacement or equal to medical advice.