|Stephen Ramsey||Caitlin Condon||Delvan Neville|
In part one, we meet computer science Assistant Professor Stephen Ramsey, who uses computational modeling to look deep into the human genome. His goal: predict who among us is more susceptible to heart disease. The information could determine not only who will benefit most from preventive action, but may even lead to new drugs for treating cardiovascular disease.
We then shift from humans to the world of plants and animals with Ph.D. student Caitlin Condon and Ph.D. candidate Delvan Neville from the Radioecology Research Group in the School of Nuclear Science and Engineering. They are both creating voxel phantom models: Caitlin for pine trees, and Delvan for marine life. By creating these 3-D models, they hope to begin to much more accurately understand radiation dose rates in biota.
[game show music]
STEVE FRANDZEL: We’re back,
I’m Steve Frandzel, and welcome back to the hit game show, Big Questions! Our last big question today is about DNA in the universe.
[crowd oohing, chimes]
Get ready, for the big prize, here we go:
If all the DNA strands in every cell in the human body were uncoiled and set end to end, how far would it reach from the Earth?
If you know the answer just give us a call at 541-7…
Oh, it looks we’ve run out of time.
Tune into our next show to learn the answer, and it’s a big one. But don’t go away, here are some exciting words from Oregon State University!
NARRATOR: From the College of Engineering at Oregon State University, this is Engineering Out Loud.
FRANDZEL: We don’t have any prizes to hand out at Engineering Out Loud, but before we’re done, I promise I’ll give you the astronomical answer about the length of our DNA. It might surprise you. In the meantime, we’ll answer some interesting questions of our own in this episode, and it just so happens that Part One has a lot to do with DNA modeling and big numbers. In Part Two, we'll find out how 3-D modeling of trees and marine life can help scientists better understand the effects of radiation. Big numbers in the form of gigantic databases are the currency of scientific research. But so-called big data alone only gets you so far. There’s got to be a way to make sense of it all, and for that, researchers rely on high volume, high-speed computing, whatever their field. Those computational tools are fundamentally the same whether you’re an astrophysicist or a biologist. Just ask Stephen Ramsey. He’s been both.
STEPHEN RAMSEY: Many of the concepts and mathematical techniques that we used in physics end up popping up again in the kind of work that I do today.
FRANDZEL: Ramsey is an assistant professor of both computer science and biomedical science who made the improbable transition from astrophysics to genetics. Now, as a computational systems biologist, he develops and applies computational modeling algorithms to sift through colossal datasets in search of genetic variations that signal increased risk for cardiovascular disease.
RAMSEY: The genetics of heart disease has many contributing factors that come from many places in the genome that, together with your environment – that includes your diet and lifestyle, where you live, exposure to environmental factors – together determine whether you will develop cardiovascular disease.
FRANDZEL: Which, by the way, remains the number one killer in the U.S. and costs the country $200 billion dollars every year.
RAMSEY: But we know that genetics accounts for a significant portion of the population variation in the risk of cardiovascular disease. So if we can get a handle on identifying the parts of the genome that are responsible for variation in the disease risk, we could, in theory, improve upon risk-factor based modeling. That’s the blood pressure and BMI and cholesterol level type of model, to one that incorporates your genetics to more precisely predict your risk of disease.
FRANDZEL: The ability to do that could have a major impact on prevention and treatment. But that’s jumping way ahead, and a lot needs to happen before practical benefits emerge from Ramsey’s research. Right now, the search is the thing.
RAMSEY: Studying how peoples’ specific genetic differences correlates with their differences in risk of developing the disease can help us pinpoint places in the genome where there seems to be a connection between genetics and risk of the disease.
FRANDZEL: To find that connection, Ramsey’s computer algorithms comb through data generated by thousands of genetic studies – each one many gigabytes in size – looking for small changes, called variants, at specific locations in the genome.
RAMSEY: So the genome is 3 billion base pairs, or letters, in length, approximately 2 percent…
FRANDZEL: Hold up, hold up. Sorry Stephen. Before we dive in, this is a good time to step back for a quick genome refresher. The human genome holds the instructions for building, well, you, and me, and those folks over there. Everybody you know. And don’t know. Almost every cell in your body carries a complete genome in the form of DNA molecules. Each set of DNA is wrapped like tightly packed noodles that would reach about 2 meters if straightened out. Base pairs are the ladder rungs of the iconic DNA double helix. They’re made from four chemicals, abbreviated as A, T, C, & G. A is always paired with T, C is always paired with G. But when, for example, an A-T base pair shows up at a specific location on the DNA molecule where a C-G pair usually shows up – or vice versa – that’s called a variant. They’re not all that uncommon. In fact, there are millions scattered throughout the genome. Genes are distinct segments of the genome that code, or provide instructions, for assembling amino acids into proteins, which perform countless critical roles in building and operating the body. Humans have about 20,000 genes that range from a few hundred to more than two million base pairs long. Variants that occur outside of genes are called non-coding variants. They’re also called causal variants, and they’re at the Center of Ramsey’s work, so…
RAMSEY: So, the genome is 3 billion base pairs, or letters, in length, and approximately 2 percent of it is the letters that actually indicate which amino acids are strung together to make a protein. The rest, 98%, is non-coding DNA. So it’s sort of like the dark matter of the genome, because we don’t fully understand what it does, but genetic epidemiology tells us that it is functional and does affect traits like height and weight and risk of disease – is totally outside of genes. So these are the intergenic, non-coding parts of the genome.
FRANDZEL: It’s in those numerically vast intergenic regions where Ramsey hunts for those troublesome non-coding variants.
RAMSEY: The general problem that I’m trying to solve is to improve computational methods for identifying the specific places in the human genome that, while they don’t code for specific changes in proteins – they can change how and when a protein might be expressed in a specific cell or tissue type or in response to a certain type of event.
FRANDZEL: So those functional non-coding variants, alone or in combination, can subtly alter and disrupt cellular communication pathways that change how proteins are assembled, how they change, and how they’re regulated in ways that increase heart disease risk. The trick is finding, out of the millions of variants, the ones that are functional – the ones that actually do something.
RAMSEY: There are probably tens of thousands of them, at least, that are present in the genome, maybe hundreds of thousands. It’s sort of like the dark matter. We believe they’re out there and we’re trying to design better computational methods to find them, but we can only account for a small fraction of them at this point. So, It’s not possible presently to directly answer the question, well within this region, which is the variant that influences heart disease?
FRANDZEL: Yet Ramsey is cautiously optimistic and believes that, eventually, genetic analysis will enable scientists to identify precisely who is at elevated risk for heart disease. That information offers a chance to take action, before it’s too late.
RAMSEY: So if someone has more compelling genetics-based evidence that they are at risk for cardiovascular disease, that may make it more likely that they could implement lifestyle, dietary, exercise changes. It may be prudent to surveil more frequently those individuals for whether they are developing coronary artery disease.
FRANDZEL: Identifying who would most benefit from early lifestyle changes is just the beginning.
RAMSEY: We have a lot of techniques that we can use to go in with a magnifying lens and look at that variant in context in the DNA. What’s around it? What proteins might be binding to that specific location of DNA in the genome, and working through that DNA location in the genome to influence the expression of other genes in liver, in fat, or in the blood, that might ultimately affect your risk of getting a heart attack.
FRANDZEL: Insights like that could become stepping stones to major medical breakthroughs.
RAMSEY: Even though the variants may not exert a big effect on risk of the disease by themselves, they can point us in the direction of cell signaling pathways that we can target with drugs to exert a big effect on the disease. The first step is being able to find those locations, then we can figure out what they do and how they do it.
FRANDZEL: Ramsey and his colleagues may not yet have reached their own big answer, but with the help of computational modeling, they’re well on their way and they’ve already expanded our understanding of genetics and heart disease. And before we move on, here’s the answer I promised: The combined length of every strand of DNA in the human body would stretch about 10 billion kilometers, or 6,230,000,000 miles – enough to reach from Earth to Pluto and back –– a long way indeed, and a little tidbit to spark the imagination about what we’re made of. Stick around for the second half of the episode, where we jump into the world of voxel phantoms. Cue mysterious music.
[Song fades into wind rustling through leaves]
CAITLIN CONDON: When I do talk to five-year-olds and they ask me what I do, I tell them that I'm just like The Lorax, and I go in, and I look at the trees.
MOVIE CLIP (The Lorax, 2012): and who are you? Who, hey, what? I’m the Lorax, guardian of the forest, I speak for the trees.
CONDON: My name is Caitlin Condon, I'm a Radioecology Ph.D. student in the School of Nuclear Science & Engineering, at Oregon State University.
DELVAN NEVILLE: Uh, my name is Delvan Neville, I'm a Radioecology Ph.D. student in the School of Nuclear Science & Engineering, at Oregon State University.
JENS ODEGAARD: And I’m Jens Odegaard from the College of Engineering at Oregon State. You’re listening to Engineering out Loud. Caitlin and Delvan are part of the Radiecology Research Group in the College of Engineering. According to Merriam Webster, radioecology is “the study of the effects of radiation and radioactive substances on ecological communities.” While that is a perfectly suitable definition, radioecology is much more fascinating than it might at first sound.
NEVILLE: Just coming in from just understanding the radiation side of things, you've gotta understand special relativity and things that happen at a quantum level. And then, to be a health physicist, then you also need to have a good understanding of chemistry and of biology, as well as the relativistic and particle physics stuff. And then, when you move the environmental side of things, now, there's all these earth, ocean, and atmospheric sciences that, are themselves a field on their own that people can dedicate their life to learning a specific niche of that, and you've gotta wrap that in as well, and-- that, for me, is one of the things that's so exhilarating about it, is that you kinda have to be a sort of Renaissance-person, as far as the the sciences, to really get into the field.
ODEGAARD: This multifaceted scientific discipline has direct impacts on the world you and I live in. Yet, worldwide the number of radioecology researchers is very small. We’re talking like 300 to 1,000 people total. In fact, Oregon State is one of only about three universities in the whole U.S. with dedicated radioecology research groups—the others are Colorado State and Clemson. The better we understand the effects of radiation on ecosystems, the better the outlook is both for us and everything else. Both Caitlin and Delvan are focusing their radioecology research on building voxel phantom models to more accurately calculate dosimetry, or radiation dose rates, on plant and animal life. Caitlin is building voxel phantoms
[wind blowing through pine trees]
for pine trees,while Delvan is doing the same
[waves gently breaking]
for marine life. We’ll circle back to more specifics on voxel phantom models later on, but first, back to the importance of radioecology and protecting ecosystems.
CONDON: I love trees. I do. They smell amazing, they're tall, they live for thousands of years, they're the pillars of ecosystems, they're important for everything-- I mean, if you like to breathe, you like trees, because we need oxygen. They hold soil together, so that it doesn't roll off of mountains and mess up the streams. I mean, trees are crazy important, even historically. The way we’ve settled across the planet has been dictated, in part, by where we've had forests available.
NEVILLE: Yeah, we had, when we were getting ready to teach some of the students about ecology, one of the discussions we had has a concept called a "keystone species," which is known-- it's an organism that plays a relatively high importance in its role in its ecosystem, relative to how much of the ecosystem is actually made up by it. So, starfish in the coastal, inter-tidal sort of zone. And, we were talking about whether or not are a keystone species, and they're kind of beyond that, because, if you remove the trees from the forest, it's not even a forest anymore. Like, they define the ecosystem that they're in, so, it's kind of important that we understand what's going on with them.
ODEGAARD: This beyond-keystone importance of trees is what lead Caitlin to make them, specifically pine trees, the focus of her voxel-phantom modeling.
CONDON: Really what it is is I'm building a more complicated computer model that more accurately defines the space and geometry of a pine tree compared to what we use now, which is important when looking at radiation in the environment—to understand exactly what the dose is. So, for pine trees, this is important because the model we have now is really just a cylinder with an ellipsoid on top, but in reality, pine trees are very different. All the living tissue is right below the bark, and all the inner part of the trunk is dead material, so what I'm doing is really building this new model with a better understanding of the living mass of a pine tree, so it's really the difference between very large mass and a very small mass, like an elephant
[elephant trumpeting, mouse squeaking]
and a mouse, kind of situation. So, with this new model, I hope to help clear up some of the questions about pine tree dosimetry.
ODEGAARD: Okay, how about you Del?
NEVILLE: Uh, so, I have some overlap with Caitlin's work except that it's with marine life, so the main question I'm trying to answer is: what is like the typical, annual radiation dose to marine life off of our coast. So, in the northern Californian current. And a big part of that comes with trying to come up with realistic 3-D models to do these calculations, because the, uh, the accepted model that's been used for these calculations, similar to how for Caitlin, the tree's just a ball-and-stick kind of thing, every -- all the fish, the crabs, they're just footballs made out of human muscle. And they go, "boom! There you go. That's a fish, that's a crab." That's -- anything in the ocean, you just make the football bigger or smaller.
ODEGAARD: These current models, think tubes and football shapes to represent very complex plant and animal life, are very generic. What Caitlin and Delvan are doing is taking advances in modeling techniques that have been applied to humans and translating that over to plants and animal life.
CONDON: So, we do human dosimetry now with voxel phantom models. So we're just applying the same techniques we're using for human dosimetry now to dosimetry for biota.
ODEGAARD: As you might imagine, making a more complex model, is much more complicated.
CONDON: So if you did it for a human, you would go in to the doctor's office, and you would get a CT scan, which is basically them taking a ton of slices -- images-- of your body. So, for you, it'd be like five hundred slices from the top of your head to your feet, and then you put them into a program where, for every single slice/image, you look at everything that's in there, so if they took a slice, right, like in your chest, you'd say, "OK, this little circle is your
heart, this little circle is your
lungs," and then what you do, is your stack all the slices on top of each other, and all those line up, and you build this 3-D image of all of the organs, and so we do the exact same thing for biota. So, for me, I go into a million scans and I circle a million pine needles, and all the root systems, and then you stack all these images up, and you get a 3-D picture from all your little tiny circles. Um, and then from there, we voxelize it, which means that we put that 3-D geometry into this kind of point-space, so it gives them all locations.
NEVILLE: So, the model that we get out, and why we call it a "voxel phantom," and -- totally accurate description, but I think maybe from a more, uh, general public way of understanding, the three-model, when you're done, is made of a bunch of cubes. So, like if you've played Minecraft, imagine if you built a 3-D model of an animal in Minecraft [minecraft sound effect?], that's basically what these voxel phantoms look like; we're just using much smaller squares so we can better approximate the animal.
ODEGAARD: For Delvan’s research, he used a micro CT scanner at Friday Harbor Labs to scan 26 marine species including lantern fish, pink shrimp, smelt, and herring. Caitlin, on the other hand, used a CT scanner at the veterinary school at Oregon State to scan four separate sections of a pine tree: the root section, stump section branch section, and small branch section--to build four separate models that she’ll then merge into one model. After creating these models, the question is “What do they do with them? Caitlin and Delvan plug their models into a software program called Monte Carlo N Particle or MCNP, which is capable of simulating thousands or millions of scenarios. What MCNP does in this instance is simulate the movement of radioactive particles within pine trees or marine life to help calculate accurate radiation dose rates.
NEVILLE: The program that we use, then, to figure out doses, it knows for whatever elements are there and what density they're at, and the energy of the radiation, what the chances are of definite interactions happening. And so, the way that we figure out the overall dose, is we have it take one particle, pick a random direction, and then roll randomly on all those probabilities until the particle has expended all its energy and it's gone. And then do it again, and then do it again, and if you do that literally hundreds of millions or billions of times, you'll come out with the average doses from a source, even though each one of those is really just following an individual particle.
ODEGAARD: After creating these voxel phantom models to more accurately calculate does rates, they can then compare these models with existing ones to see which would be better to use in a given situation. Delvan adds that beyond just comparing models, another question arises: can these new voxel phantom models be scaled to other species?
NEVILLE: You can't always make a 3-D voxel model for every specific organism in an environment. If you don't have a good 3-D model, but there's a similar 3-D model, can you just take that model and scale it by dimensions to approximate the organisms you're really looking at? Or, does that just not translate?
ODEGAARD: The answer so far is still up in the air, but at the end of the day, these models for now will help Delvan, and Caitlin and other radioecologists, more accurately understand dosimetry in plant and animal life.
CONDON: There are still a lot of things that we don't know, and so it's really important that we develop these models and we are working on understanding dosimetry to biota. And, when we understand dosimetry to biota, we can start understanding what's happening. there are places in the world that have been contaminated from nuclear disasters, like Fukushima and Chernobyl. They are our existing sites with radiation contamination, and we can see, um, that there are some effects, but there's a lot of misconceptions and not really understanding what exactly the effects are to biota, and so really understanding dosimetry will help us really understand what the actual effects to biota are from radiation.
ODEGAARD: This challenge of uncovering what is currently unknown is the driving force for Caitlin and Delvan as they expand the collective understanding of the whole radioecology field.
CONDON: Yeah, it's kinda beautiful that way- I mean, I don't know, I'm sure there are tons of interdisciplinary fields, but radioecology especially is such a small field, and it's so interdisciplinary, that you just have to get really comfortable getting out of your comfort zone and admitting that you don't know a lot of things, which I think it really important. Like, you can't be a radioecologist if you're gonna go around pretending you know the answers, because you don't. If you're really into science and wanna learn a lot, it is a great field to be in.
NEVILLE: I enjoy it, because it's one of those-- I mean, it's like this across radioecology, but there's so many things you can find where nobody looked at it yet, where you can just-- it's the first time anybody's even really tried to figure that out, there's so many unknowns.
ODEGAARD: With their research, Delvan and Caitlin, are helping eliminate these unknowns one Minecraft-like voxel phantom model at a time.
This episode was produced by Steve Frandzel and myself. Final audio editing was provided by Miriah Fordham. Our intro music is The Ether Bunny by Eyes Closed Audio on Soundcloud. We used it with permission of a Creative Commons 3.0 license. For more episodes and to subscribe, visit engineeringoutloud.oregonstate.edu, or search “Engineering out Loud” on iTunes or your favorite podcast app.
MIRIAH: It's Miriah Reddington.
ODEGAARD: Oh, fine. Final audio editing by Miriah Reddington.