 Welcome again to the Martin Jiski Hall of Biomedical Engineering, to the panel discussion as part of the Purdue Engineering Distinguished Lecture Series. I'd like to introduce to you our moderator today, Professor Chris Rocher. Professor Rocher is in the Medicinal Chemistry and Molecular Pharmacology Department and is Director of the Purdue Institute for Integrative Neuroscience. He's a well-recognized expert in protein misassembly and aggregation in neurodegenerative diseases, such as Parkinson's disease. So we're delighted to have Professor Rocher moderate and introduce the panel. Please welcome Chris Rocher. All right. Well, thank you, George, and good morning, everyone. It's really an honor to participate on this panel discussion, and really I'm excited about the opportunity to further explore the topics that we've just heard about in this forum. I think the goal here for all of us is for this to be really a fluid discussion. We do have some questions to get us started. We have some questions submitted by the students as well, and so we'd like to make time for those. And at the end, we'll certainly allow some time for the audience to ask questions. So just to begin, I will have the other panel members introduce themselves with just a few lines about your background and expertise as it relates to what we'll be discussing here. So my name is Fanghua, I'm an assistant professor at the biomedical engineering department. So my lab work on a single molecule imaging, just as you have heard from Dr. Murner, and super resolution microscopy. And we work on the direction of lifestyle imaging in tissues and hopefully animals. And then we push on the resolution that we can achieve in 3D. I'm Tamara Kinzer-Eartham, I'm an associate professor in the Walden School of Biomedical Engineering here at Purdue. My lab studies biomolecule networks, particularly protein signaling networks, where we look at them as dynamical systems and try and understand how their spatial localization and gradient activity inform cellular decisions. And particularly, we're looking at synapses in neurons and studying kind of learning and memory formation. You've already been introduced to Dr. Murner, so we'll move on to Garth. All right, I'm Garth Simpson, I'm a professor in chemistry. We build instrumentation using ultrafast objects to probe nonlinear optical interactions as a function of position and time. Dr. Garth, we work on two aspects of superresolution, one relates to fluorescence imaging as an extension of the way we found to image frame parameters in vivo, but instead of using the point spray function of the microscope, we can think about a transport problem and then the point measurement, the Green's function. And the coherent aspect of superresolution is when there's a structured elimination, especially in the field, you can think about the object moving in the background or the field moving in relation to the object. This encodes far beyond information in a way to extract super resolution information in a way that's fully yet to be explored, hopefully at least. All right, thank you, everyone. So I think we have a sense that there are people of interest from a fundamental perspective and applications. And so I'll ask the first question just to get going. Dr. Murner, your journey is really fascinating, what you've shown us. I think you did a tremendous job illustrating your path. I think one question that many of us have is how, where were you at the magnitude of what you were doing? If you look back, do the various steps make sense? Were they logical? Did you realize that at the time? Or how much of it was serendipity? Yeah, that's a great question. Thank you. It turns out that the steps that we were taking make some sense, if you like. That's not me making all that noise. They make some sense in the sense that my life was going. That with starting out as an electrical engineer and then learning about physics and mathematics and chemistry and biology and so on, I mean the time at time that sort of asking these questions that were fundamental questions was the most important thing. So let's put it this way. We knew that it was going to be exciting to detect a single molecule because no one had done it. So it's like a tough experiment, right? Or something like that that you could say, well, we did something tough for something. But that's of course not impact, right? And the impact that occurred, we could not envision. Remember we started at low temperature, you might say fairly esoteric. It required high resolution and lasers and so forth. Liquid helium superfluid. But by opening a door, what is so exciting is that the scientific world, the scientific community, everybody thinking about that. New people doing experiments, those stimulating us. Our work stimulating other people. That all happens and it's a bubbling thing that keeps going forward. It wasn't clear at the beginning that blinking would have an application, for example. And yet it does later. So this is really the way science works. You have to be excited about science every day. You have to be excited about what you're doing now. You have to enjoy it all along the way. Because impact didn't mainly not occur. But the best thing is to be happy, which is to enjoy it every day. So I was following this latter rule. Not saying, okay, we're going to do something. We're going to get an Nobel Prize someday. So we're going to do only something that will give us an Nobel Prize. That's just not the way to do it. I guess another question many of us have is then, what is the next big thing to lead to the next revolution from your perspective? Or what are some barriers that you see right now that must be overcome? You see, that's another good example. People would like to know what, have me tell them what's the next great thing that they should do. So I've done my best by telling you things that were just finished and just submitted, right? The current challenges, of course, are to always learn more. Always extract more information. Always extract more useful information. More useful variables. Try to figure out ways to measure more properties of the system. And we're choosing to do it with single molecules. Not everybody uses single molecules. But just because they're this sort of elemental, smallest unit, I still think it's a very valuable and fruitful area to explore. Because there's more to learn from that regime. And putting it in a different way, there's thousands and thousands of experiments that contribute to a whole text on cell biology. And we've only reproduced a few of them so far. With super-revolution and single molecules. We should reproduce all of them. We should reproduce all of them. Good try. I think it's coming in and out. Can you hear me? So, can you hear me now? Four, five, six. I mean, the people that are local can hear my voice anyway in a room like this. So we don't need all these microphones. And if it works okay for the distance, fine. So anyway, I think that's a major frontier. In my mind, work that needs to be done is on fluorophores. We need better switchable fluorophores. We need fluorophores where you have more control over whether they're on or off. Turns out they all behave with Poisson statistics, which is not the best when you have an exponential distribution. You'd like to actually have a narrower distribution when you want to change a molecule, for example. You have lots of needs for different wavelengths. We have needs for better fluorophores at 77 Kelvin that can be switched on and switched off. So, you know, most of these things are incremental. More photons, better precision, flying in more systems and so forth. That's not really a revolution, but some of those things may lead to an observation which then does produce a revolution. That's where the revolutions come from. Just doing great science, good science, fundamental science. Not predictable. Okay, let's have you just pass the mic. That one's the one that works. Thank you. Okay. Since I don't have the mic, I'll go ahead and ask a question. Sorry, he's a moderate. No, I think I was hoping the other panel participants would chime in at this point. I'm really curious how you made this initial transition from doing spectral hole-burning measurements on defects in crystals to single molecule detection. And then since then, how you managed to not only understand the physics and the mathematics behind it, but also the biology and the implications of the biochemistry associated with these discoveries. So, you see, all of these transitions occur because I'm always trying to think about what's interesting, what's exciting to do, what's the next thing to do. How can I have more impact, if you like? So, low temperature was high resolution, as you said, and in fact, high resolution is really beautiful. You know, when you can see many, many, many features just by scanning the wavelength of the laser and all sorts of beautiful things, and it was very sensitive to the local environment, all kinds of things, and it was hard for me to leave that regime. But remember at IBM, I couldn't do the biology. And so, leaving IBM opened up the possibility to apply to biological systems, and because single molecules innately, since heterogeneity, okay, if you measure single, single, single, single, I felt it would be much better to shift to the broader biological regime and give up on the super high resolution, which is very, very difficult to do. But you can see what happened. So, because of the breadth of applications and the breadth of systems that can be studied and the problems that are present, you know, at room temperature, that turned out to be a fruitful step. So, it's just a matter of at certain times in your life, you have to make a choice. You have to decide where do you want to go next, and basically remember that all the things that you've learned, this is to the students, of course. You learn many things and you stop doing one thing and then do another thing and so on and so on, and don't be upset about that. Because all of those things you did before, you might use later. So, think of it as a process of continually learning new things. And that's anyway what my life is all about. I'm continuously a student. So, learning the biology was another challenge. You have to be comfortable a bit with being a beginner in a new area and bone up on things quite a lot. And so on. So, it's challenging. Let's put it this way, but I encourage challenges, right? Challenges are actually better for you than doing something that's boring. I had to smile during this morning's comments, because I think that my personal opinion is fundamental understanding and pursuit of interesting directions with passion is really important. And why I was particularly smiling is that as a gentleman at the end of the row here, who helped me understand some... We'll start to understand some things about the brain. And that took quite a lot of patience to encourage me to enter the field of neuroscience a few years ago. I... So, I was also thinking about the sort of the fundamental... What people learn, how they think about problems, whether it's interpreting an experiment or thinking about basic math and physics and how you sort of consolidate and understanding and knowledge as opposed to just information and you build upon that. I have a technical question, but maybe I should wait on... Go ahead, I think. Please. So, super resolution comes from adding information that you know about the experiment, the point spread function or the fact that you have something blinking. So ultimately, at the end of the day, it's a signals to noise question. How many photons or how patient you are or how stable things are. So there's a lot of practical limits. Then you have fluorophores that are introduced, targeted, and they blink and you see them to high resolution. So you know where they are, at least at a certain instance in time, and you... But the information that you can extract about the underlying questions relates to where they are and what they're doing and how they're interfering with the system. They're reporters on something that maybe you care about as an underlying mechanism. So every measure interferes to some degree. You have to shine photons on something to see. So there's a question of extraction of information and underlying principles, I suppose. So maybe I should mention two things. About A, it's ultimately a difficulty question then in terms of precision, signal to noise, B, about what you actually learned about the underlying problems and the strategies for that. I think you want me to comment. Sorry, take this back. Sorry, yes. So those are very useful points. One of them that you have mentioned a bit here is this issue of attaching a fluorescent label or something like that and the effect of attaching a fluorescent label. So there are a couple of things to think about when you think of that particular issue. First of all, it's always a fair question to ask whether you're perturbing the system. I mean, it's absolutely essential that we don't just measure something that's an artifact or something like that. Luckily, let's talk about green fluorescent protein for a moment, it would not have revolutionized biology if it perturbed systems terribly. So the Nobel Prize in green fluorescent protein to Roger Chan and his colleagues and so on in the field are resting on the fact that that's not that great of a perturbation to add that protein for the most part. But I'm not going to say it's always true that there's no perturbation. But if you think about it from the point of view of something that engineers study very well, we're working at low Reynolds number. We're working in these situations where the fact that you add that mass doesn't matter as much, okay? Because it's not ballistic motion. It's not ballistic transport. It's diffusive motion, okay? So that has mean square displacement growing linearly with time. And so there aren't velocities and so on in the sense that you'd normally think about where you might worry about that kind of perturbation. The other aspect that comes to mind when you think about these issues, so the comment I just made relates to whether the motion is perturbed a lot by having to drag a floor for around with it. Well, if it's a green fluorescent protein, that's of course a lot bigger than a single 1 nanometer label. And so whenever we can get to the single 1 nanometer labels, that's obviously better. You get more photons and it's less perturbed if for sure. Anyway, there's such ease in doing fluorescent protein labeling that people still continue to do it a lot because it's so quick and easy to get that kind of labeling. And I want to say that now we know that for certain structures that fluorescent protein label does perturb the super-resolution structure in certain cases. If you, for example, over-express GFP fused to a protein called CRIES, which is a crescentin-like analog protein and colobacter responsible for its crescent-like shape, that protein is supposed to just be in a crescent line across along the edge of the cell. But when it's fluorescently labeled, it can detach and form a helix, which is not in the normal shape. And so we know that. And the best thing to do is to test a structure with several different labels of different sizes to make sure there isn't perturbation. So I guess what I want to say is, because we have super-resolution now, we have to be even more careful because now we're going to possibly begin to see cases of super-resolution, right? But nevertheless, remember that there's other objects that one can study, a natively fluorescent protein. The antenna proteins are common in photosynthesis and so on. You don't have to add any external label. We study those a lot, even in solution. They have fascinating photodynamics. So it's a great question. You always want to know if there's a perturbation and you have to do controls and you really have to do controls and sometimes physical scientists are not so aware of that particular aspect. But it is only by great controls that biology made progress based on two-by-two correlation matrices, plus, minus, plus, minus. Only by doing a lot of controls was that able to make sense. Sorry for the long answer. I would riff off of what you were saying earlier about switching fields. And the way you can switch fields is either you learn it yourself or you bring in collaborators that are really good in that field. And what struck me about looking at some of your work is your collaborations with Lucy Shapiro on the Colobacter and the work with the glycocalyx with Caroline Betosi. And that's allowing you to look at the glycocalyx without fluorescent-linked labing with genetic proteins and using some of these bio-arthogonal chemistries that are coming in from chemical biology. So again, kind of crossing fields and allowing you to put labels on the outer matrix of the glycocalyx without using these really large genetically encoded proteins. And that's another switch in fields. And I think that technology in particular has a lot of interesting applications moving forward. So you're absolutely right. I wanted to mention earlier, but again, my answer was already too long. The importance of collaborators, for me, was really essential for this changing field business because the stance we take is that we're expert on the physical measurement, the optics, the single molecules, localizations, and all that stuff, the optics. But I like to collaborate with good experts on the other side who bring all of those other knowledge and skills and background to the problem. But it's super important to recognize what makes a great collaboration. A great collaboration is where there's mutual interchange of ideas. And you learn about the other person's problems and they learn about your problems and you know both and work together on a solution. So what does not work is throwing something over the fence, I call it. You know, oh well, we're the people that do one thing and so we make a measurement, we throw it over the fence, the people on the other side do their thing and throw it back over the fence. That's not a collaboration, so I've been blessed by the collaborators, like you mentioned several. And others, Peter Jackson, an experiment looking at the Inverson compartment of the psyllium, you know, Tim Stearns, working on other structures in the psyllium in the centriole and so forth and so forth. So there's been a wonderful flowering and I think that's a great way to learn new fields. So if I could pick up on that a little bit. I think that what you're saying really resonates with many of us here. I'm really focused in the Life Sciences Institutes, including the Neuroscience Institute in which I'm involved with trying to spur collaborations between life scientists and engineers. And Kevin actually alluded to it earlier. Sometimes it's challenging, we speak different languages, but I think essentially all of us involved in these collaborations find that they're really enriching experiences. I'm wondering what you think though in terms of how that influences training. You know, I think many of us think we can provide different opportunities to our students as a result of having these collaborations, but we have to think carefully about what is the most productive experience for students in those settings. So I'm wondering if you've had similar thoughts with your collaborations venturing into different fields from your immediate field initially. Yeah, it's a great question and it's worth thinking about. I still believe it's essential to become expert in a particular area. I mean, when you're, let's say, getting a PhD, you certainly want to be expert in that area. You don't want to just be jumping around and never getting the depth to be able to execute something deep and complicated. But the fact that there's these collaborations often certainly gives opportunities for cross-fertilization and learning of the type I just mentioned in terms of learning the other side of the story. So quite often my collaborations with biologists have involved one of my students working with one of their students or my postdoc working with their postdoc together and so on. And so they, that certainly gives both sides a more powerful knowledge set for the way we should move into the future. You need to be working and understand what's going on at the boundaries of fields to make progress. I just, I want to chip you in a little bit. I'm still learning this, but we did exactly what W has mentioned before, serving expenses, right? This is the initial error said when we're trying to do collaboration. We thought we could collect data and the biologist is going to interpret the data. But in fact that this doesn't work so well and we have hard experience on that. So what we feel these days is if we want to train a student to work on a collaborative project instead of getting a student to go to individual meetings we set up collaboration of meetings and then we have to review the progress at least every two weeks and discuss and what's the bottleneck. And then hopefully a range of people can sit in the meetings. I feel at least when we start from there and there's a lot of involvement and ideas and as a PI I also learn a lot during those meetings. Maybe I can just ask one more question from a technical point of view. So you mentioned your BGNet, right? This is a fantastic demonstration of how deep learning can be useful. And the pattern background that you mentioned before it has been always there, right? In super resolution imaging that we see single molecules there there's always background. The background never is constant. They are always patterned. But nobody solved that problem previously. So first question is why people didn't even attack that problem. The second question might be maybe what do you think deep learning can go further with, for example, optical imaging and super resolution in general? So yeah, those are great questions. First of all, in terms of background you're absolutely right that many researchers have used just a constant background in their fitting. But there are some researchers who have pushed this one step further and used particular kinds of ways to estimate the background. An example is rolling ball estimation of background. So this basically means you have this lumpy image that has a bunch of molecules in it but you roll a ball underneath the image and whenever it touches the signal then you call that the background. And that actually works but only for low spatial frequencies. It only works for a certain small range of spatial frequencies. So my point is that there are some people who have tried to do things and done useful things. Another example of that might be my single molecules are very sharp points and so I take a Fourier transform and I look for, as the high frequency information is where my molecules are and the low frequency information and everything else and I use that and then inverse Fourier transform and then fit. So for those kinds of tricks have been utilized and so why has nobody tried to attack it in this fundamental or sort of broad way? Well, I don't know what to say about that. I mean, Landhart is a brilliant guy. So he has decided he wanted to learn a lot about neural nets and taught himself. We're seeing it affect our world in many ways so it's not such a bad idea to learn something about it and there have been several other neural net applications he's come up with. We're using it for phase retrieval, we're using it for some fitting, other people are using it for multi-emitter problems and so on. If it's stated as a clear problem that can be solved by managing, using this as a way to refine weights in a very high dimensional space then that's not so bad. It sort of lets us solve maybe a problem that's harder to do from pure fundamental points of view. So I don't know, I think it'll keep going and have more applications but we always have to look at them closely. Can I ask a follow-up on that? So in super resolution methods you are detecting the 404 and when you start labeling and you're labeling density therefore dictates the resolution of the structure you can obtain. In other words you can only see something, a structure if it has 404 dotted all along it and at some point as we increase the resolution the labeling density starts to be the limiting factor. Do neural networks for example create opportunities for filling those gaps in super resolution imaging? Well, I mean the latter part of your question is not so clear. I mean maybe, I would say probably but more specifically this issue says, okay, we're doing labeling we're measuring positions of molecules but if you don't have sufficient density of labeling as you said you cannot claim high resolution and for that reason some people in the field have defined new measures of resolution not just the localization precision or something like that but a combination of localization precision and the density of localizations that you find it's so-called Fourier ring correlation it's a different measure that includes this effect, okay so you're right that if you want to get higher and higher and higher resolution you do need denser and denser labels then they'll start interacting with one another and that'll be an ultimate limit that'll be a problem that has to be dealt with, right whether neural nets can solve it or get around some of these issues of density and so on is not clear but I think there's already some progress in this area and some people have just recently published some papers that say well if I only get sparse localizations I can figure out what the structure is but they already know what the structure is okay so that works where they knew what the structure is supposed to be and if they only sampled it poorly then they can figure out where the rest of it is okay but you can see where that came from it's from the prior knowledge of the shape so there's lots of interesting challenges I would say and that's one of them at this point I'd like to so yeah I'd like to cover some of the questions that were submitted by the graduate students because these are good questions I think will be interesting for all of us here so one question is that a number of studies in the field of super resolution microscopy have focused on single cells bacteria or mammalian cell culture and so the question is what challenges do you see in moving to tissue level studies so and tissues of course there's a number of issues certainly one of the big challenges is that they're a lot thicker and you have potential for out of focus background so it's very important to have use some imaging method that rejects out of focus background you could say that might be light sheet but you also could say let's use something like STED stimulated emission depletion microscopy which has a confocal sectioning capability we actually use STED in looking at tissues right now in the lab it's a super resolution technique invented by Stefan Hell it's great but very complicated compared to single molecules but nevertheless it does have some Z sectioning so we use it on tissues the going very very deep in tissues and so on gets into lots more complicated and fascinating optical problems scattering and all kinds of other things and there's a lot of smart people in the engineering community trying to figure out a lot of loss of information if the photons scatter a lot on the way out I'm excited to learn what's going to happen in that area but anyway in general you can use it on larger structures you of course want to be able to look at a large area in order to find the area of interest because you're always looking at a very small region when you're doing super resolution yet where the region you care about is somewhere else far away from that and you have to be able to scan everything before you zero in things like that but I don't see a fundamental reason except for these issues depth, how do you deal with depth sufficiently, how do you deal with scattering how do you deal with out of focus background and those are challenges so another question again on the application side relates to how do you imagine the evolution of technologies advances in the area of super resolution with electron microscopy I think you give a good sense of those two working hand in hand in your talk so what do you foresee in the future as far as those two approaches yeah well the advances in electron microscopy right now are really staggering they're really amazing just by improving the textures and using some very simple phase mask actually zone plate they've been able to improve their resolution tremendously but in the broader cell context you don't get the same resolution as you get when you're trying to look only at one protein and average together hundreds of thousands of copies of that one protein so there's limits and there's a limit on the thickness of samples so people know that and so they're already using milling and so on to slice samples and get a bunch of thin sections they require thin sections you damage the sample of electrons when you get sufficiently high resolution and you have to trade off again all these advantages and disadvantages I'm sure that some people in that field think that they're going to ultimately have so much resolution just imagine improving it by another factor of 10 or a factor of 100 who knows how you're not changing the wavelength very much your detectors are already spectacular anyway, that's suppose they did a factor of 10 or 100 then they will see the shape of every protein okay and they don't need for essence maybe okay but that's not true now so for now remember that there's value in watching individuals and seeing how they change with time and seeing how they are in live systems and seeing how they fluctuate and seeing how they might behave differently and you sort of miss that when everything is frozen advantages and disadvantages exciting another question I just wanted to make a few more comments on that so in terms of thinking about super resolution just to dive a little bit more deeply on two things that you mentioned really quickly is that just cryoem you're going to run into super resistant problems and that's what we're seeing we're doing some cryoem I see my postdoc Scott in the audience even if you're looking at a single protein and you cannot tell what it's 3D structure is if you're looking through a 2D plane and you need a large amount of prior information about what the structure is that you're looking for especially when you're looking at cells then and you mentioned this in your talk is you're going to need a lot of information about what you're looking for and it's very static so that would be the third point is you cannot get dynamic information from cryoelectron tomography or microscopy these are areas where super resolution microscopy and the interface between the two are going to be very exciting I think another question switching gears a little bit so we talked a little bit about your mentoring method but this is an interesting question particularly with all the fascinating advances in your lab so do you discuss with your graduate students which parts of the project they can take when they graduate how do you make those decisions there seems to be a lot at stake there well you know absolutely we discuss that when the students leave for the most part there's so many new ideas out there so many interesting things to do that this is seldom a problem or putting it a different way when people in my lab accomplish something publish and write it up and so forth it's usually pretty exciting and kind of important if possible not every paper is super exciting of course but they tend to want to do something different they tend to want to use everything as tools and switch to another area so it's really very seldom a big issue in the in the case of tools of course in general everybody can use a tool you publish the tool and everybody in the world is going to use it certainly I'll use it if I want to and that makes only common sense so it generally works out without any any big difficulty so one more question here and then we'll open it up to the audience this is another interesting question a bit on the lighter side what's your favorite part about the Nobel Prize Ceremony? these questions are always what's your favorite element so I was asked what's your favorite element at one point and I was like what so but that was at a conference the Mendelaev meeting 150th anniversary of the periodic table and so on so the the favorite part about this whole activity I guess I would have to say was the pleasure of bringing my family colleagues and friends to Stockholm and having a very special dinner of our own separate from all of the hubbub that we took them to and that was a great moment because it combined family but mentors and colleagues collaborators and so on in different parts of the whole activity the the activity is pretty crazy the whole week the whole Nobel week is completely amazing and there's long stories that I'm happy to tell about it if anybody wants to listen there's also a book about it that you can read one of the laureates wrote a nice book with King Gustav is the name of the book Reindeer with King Gustav the wife of Bob Laughlin wrote that one has a lot of the correct details if somebody's interested in it but the perhaps the least favorite was some the surprises that occur when people just contact you because they're trying to take advantage of you and so I'm being very honest here what actually happens there's people who want to somehow profit by asking the laureate to do something you know I was asked to come to an event and give talk and so forth but it turns out that this was a stage for some person to sell a bunch of things and make a bunch of money and in charge people a lot of money for coming and talking to the laureate and things like that are not so much fun when you recognize that in some sense there are a bunch of people out to make a buck out of it so you for example there's people in Stockholm so after getting out of the car usually the car is a fancy black car that has a nice limousine with a special driver and our special FSA run you know going around with this people but when you get out quite often there's a couple of people standing there and they have a portfolio and they whip out a huge picture of me and they say please sign this and I'm going oh ok that's interesting I guess I can sign this so I sign the first one then we go to the next place the same two guys are there same two guys pull out another picture one another signature so on and so on so these are on sale at ebay for Nobel laureates they sign pictures and so you're always wondering right what's somebody going to do with my signature so that's not the nicest part I think in retrospect I guess it's not surprising but I'll add one other question if I will what is your favorite element since you opened the door I think everyone wants to know you don't have to answer so everything I had to answer was very no notice I said Bismuth that's because in graduate school we were working on far infrared spectroscopy far infrared detection, far infrared sources far infrared line shapes all kinds of stuff and one of my projects was to try to take Bismuth which is a fascinating semi-metal and turn it into a light source for far infrared you know sending currents to it whatever two degrees Kelvin so anyway, Bismuth is what I said sorry you want the mic great answer alright sorry if you wanted to hear about Bismuth I think at this point it would be great to open it up to the audience do we have microphones or do we need to give you yeah so with super resolution microscopy what do you see as applications that will translate to the practice of medicine so in the case of medicine you know you care a lot about disease and but it's worth remembering that understanding of disease is based on knowing what normal behavior is as well as disease to behavior so all of these techniques apply to both that is we can learn more about cell biology when it's normal and we can learn more about cell biology when it's abnormal so the example from my talk was the Huntington aggregates and these structures and cells that come from that kind of disease but there's many other you know manifestations of disease that come from cell biology and so the real regime where this matters is because you can see things better you can also ask what is and at a mechanistic level what is a particular treatment doing what is it actually changing so that's all about super resolution but in backing up to single molecules single molecules are being used of course in sequencing now in the pack bio and so on other companies are using fluorescence from individual molecules to sequence DNA very long reads come from the pack bio implementation and so on so there's a lot of connections because it comes from at the core cell biology understanding it both in disease and normal situations this happens there's many many examples drugs can be labeled you can see what a drug does where is it going how does it get into the cell all kinds of things like that thanks for the great talk I don't know that I have an example but I would have a cautionary tale just to think about what happens in medicine in drug discovery when we rush ahead to develop drugs and put drugs out in the market when we don't understand the fundamental cellular mechanisms and we'll take the neuroscience for example as a case where we have now pushed into phase 2, phase 3 clinical trials multiple Alzheimer's disorder therapeutics and they have all failed because we still do not understand the fundamental mechanism behind the disorder so if we look at fundamentals and put our energies in understanding that basic science ultimately will be more successful in developing therapeutics and treatments for disease absolutely need more of that back to Fundaman thanks for the great talk I have a question about the neural network just like by the neural network we can go beyond the optical limit and get some hidden information from the image but how do we tell the information we got is fair or some artifact from the neural network processing yeah that's a great question absolutely and very very important so the best answer that I have is you must validate you have to validate many many times it's the equivalent of doing controls you have to do enough tests on known systems and see that it behaves correctly on known systems and trust it on systems that you may not know everything about and so that's the basic answer to that problem and that's why we train with train sufficiently so that you get the right answers back most of the time not all of the time there will be a few cases where there's a few estimations let's say of the background that are incorrect so it only comes from having this statistical weight of showing that many many many cases hopefully sampling the space very deeply you see that it behaves correctly I have a question what is the dynamic in your lab with you and your postdoc and your PhD students for example if your PhD or postdoc encounter a problem do they always come to you to find the solutions where you let them to solve the problem alone and if your postdoc have a good idea would that idea come from you to think about their own project well I would say that all of these things occur in different degrees to different people but in general I'd like to be interactive with the lab my office is in the basement with the laboratories not with a window to the outside it is because I want to be close to the research when a problem comes up sometimes they don't ask me and they try to solve it themselves and they sometimes do which of course is fantastic which is great but there are sometimes when there is a problem that they haven't solved and I actually can solve it very quickly because I've seen it before so they learn over time that it actually is not a bad idea to ask me whether I know how to solve this problem so in any case it is important for very important for grad students and postdocs to solve problems on their own and it's a tough thing I like to be involved so that I'm always asking for that if there's a new idea for a new project and I find it exciting and our funding structure can support it then I'll be happy to encourage somebody to follow that so the tilted light sheet idea was my idea actually but other people implemented it, made it work, showed it fixed it, that is there are always going to be details that require things to be adjusted and made to work better and so on so it's really still a team effort in the end so some ideas come from me and a lot come from the students it's a mixture and I'd mentioned by the way that whether funding will allow it so unfortunately this is a kind of a bit of a fact of life in our world we have agencies that we ask for money and then we have to give reports and then we have to get refunded and we have to have the money to keep the lab going so it's very important to actually deliver on some of the things we said we were going to do otherwise you don't get funded and I've had grants turned down just like everybody else so it is important to also think of the funding structure sorry that's the way it is so it seems that structured background noise can be difficult to analyze for neural networks I was curious what steps can be taken to reduce the incidents of these like figures well I mean I showed a system that actually extracts structured background it estimates it quite well so it's easy to remove but you can also experimentally try to reduce it for example it's always essential in any of these experiments to have the cleanest possible situation to remove any impurities fluorescent impurities that might be causing background for a long time at the very beginning we grew cells and phenol red media this is just thrown in but it's not essential it's just to show you the color of the media and that actually produces fluorescence so you start growing without phenol red so you lower your background fluorescence but native native auto fluorescence from the cell is hard to get rid of it can be something that can cause difficulties so you can move to longer wavelengths to move away from the background fluorescence that's from auto fluorescence that's why we do a lot of imaging in the red or yellow to red or whatever because you move away from auto fluorescence trying to do these experiments in the blue is extremely difficult and almost no one does it because of extreme background from everything that starts lighting up right and then you can also use maybe time domain if you were clever you could might figure out a way to throw away background because it maybe has a different time structure than the emission that you care about this is a technique called photon burst detection and so on so there's ways to think about it but nevertheless sometimes you can't get rid of it easily and having a mathematical way to remove it is very convenient thank you so much I believe from your talk you already showed like from the single molecule like a perspective the technology or the method here is very not fully developed but well developed but in some of the cases for example in Alzheimer's disease and some like a cancer biology people are more interesting to understand the protein-protein interactions since that kind of dominates the fundamental mechanisms behind the diseases do you think how far we are or what are the challenges we're facing right now moving from single molecule imaging to two bodies or even three bodies observations using super microscopy super resolutions well so we do a number of protein-protein interaction studies that involve using multiple colors placing different objects at different colors and then you can watch them simultaneously and FRET has another technique it works on a very short spatial scale two color co-localization works for intermediate scales and so on so we already have techniques from the single molecule community for looking at protein-protein associations however one of the challenges is if you'd like to do that at the single molecule level then you need both of those to be at low concentration you need and if you take two objects both to an enomolar then they have very very low probability of interaction so one strategy that people have been using to get around that is to use these controllable floor force ideas if you leave most of them off but only turn on a few very close to the one you're interested in or have one be in high concentration but only sparsely labeled and so those techniques allow you to observe protein-protein associations I mean you know unfortunately your question opens up, it's a great question opens up a lot of interesting things to think about another way to measure protein-protein associations is to use our thing that we call the ABLE trap this device that looks at individual proteins in solution and what you can do is the trap analyzes everything about the trapped object not just its fluorescence and let's say it's excited state lifetime and all that but also the behavior of its motion how it jiggles around in solution in the trap which is information about its size and its charge and so if you can measure the size and charge of the trapped object then you can have other partners that are coming in that are unlabeled but when you form a dimer the behavior of the motion changes you form a tetramer of the motion changes and so on and we applied this recently to an oligomeric system so it's not so different from these amyloid disease kind of problems if you would like to know more precisely what's the distribution of oligomers you can do that by sensing something about the object when its partners are around so I think there's a number of ways to look at protein-protein kind of interactions some based on diffusion some based on fret optical kind of tricks and so on and there's more ideas that people have out there for this sort of thing yeah so I think the question was not, we couldn't quite hear it so he's very interested in knowing what criteria or what I think about when I'm selecting graduate students or postdocs for the lab, is that correct yeah so first of all I look for achievement in their prior work that they're an expert in a certain area I've learned a lot that who are inquisitive who are ready to ask questions ready to think about what kind of things have not been done yet I'm interested in people that have done things with their hands this is actually a tough one now because there's not that many people building stuff with their hands but I try to find out can I figure out if they're good with their hands and do they repair their bicycle do something that builds something because we're really experimental we have tables filled with optics and components and so forth that all have to be adjusted so that's another key thing that I need but in general it's all these things together plus their background does it fit in at a given time does it fit in with our funding so on all of those things hi Dr. Moore I would like to ask a question about how exactly do you deal with the heterogeneity that you encountered during cell experiment because there has been a lot of effort being pushing the temporary and spatial resolution but with such a small field of view so it takes a lot of effort to be able to collect a lot of images from different cells and usually those cells have it's an intrinsic heterogeneity with it so I would like to know what's inside about balancing the resolution and also more information about the population so it's a great question one answer this is a little bit of a flip answer biology is tough okay biology is complicated but more seriously you know if the cells are so heterogeneous that we cannot see any consistent behavior then we're just doing the wrong experiment and we have to do something else we have to have some behavior that's repetitive that the cells have to generally behave the same and your question is sort of how do you it takes time to do all this well you can program all that I mean one of my postdocs decided to do a big survey of the whole field make some quickie measurement of all the cells record where all of them are then you just tell the computer go to the first one do the measurement students at home or doing whatever data is being taken and so we fight it because you definitely do have to observe many single molecules one at a time and look for consistent behaviors it is one of the aspects that sometimes is hard for some people to get comfortable with but it's sort of a fact a fact of doing life a fact of life we don't have control of every variable so that's the challenge you're right well this has been really terrific time has flown by and unfortunately I need to bring the discussion to a close I think we could have all gone on much longer this was really excellent so I'd like to thank my fellow panelists here of course and thank the audience really terrific questions it was great for everyone to be so engaged thank you for joining me and thanking Dr. Warner for really giving us a memorable stimulating experience here