Florian Jug (Human Technopole, Milan)

This is a machine transcription and therefore it may contain inaccuracies, errors, or mispronunciations. Notice an error you think needs changing? Please contact the Bitesize Bio team using this form: https://bit.ly/bsbtranscriptions

Intro/Outro (00:00:02):
Welcome to the Microscopists Aide Size Bio podcast, posted by Peter Oall, sponsored by Zeis Microscopy. Today on the Microscopists,

Peter O'Toole (00:00:14):
Today on the Microscopists, I'm talking to Florian Jug, research group leader and head of image analysis facility at the Human Technical Milan. And we hear about his research using AI to better analyze and quantify biological data.

Florian Jug (00:00:30):
Deep learning will in, in, in general machine learning and like statistical analysis will play a very important role. And I think, yeah, I, I can't wait.

Peter O'Toole (00:00:40):
We discuss his passion for teaching courses across the globe,

Florian Jug (00:00:43):
Uh, because I'm not in a university, so I don't have to do any teaching. And if you're not forced to do something, then maybe it becomes wide. I'm not sure. The whole endeavor is built on us building useful things for people,

Peter O'Toole (00:00:57):
And we chatter about his early ambitions in computer science.

Florian Jug (00:01:01):
I was already coding at home and we were moving quite a lot. Um, so there were always, uh, uh, years where I didn't have a huge amount of friends yet, and that was great times to be friends with my computer.

Peter O'Toole (00:01:18):
All in this episode of the Microscopists. Hi, welcome to the Microscopy. I'm Peter UL from the University of York, and today I'm joined by Florian Jug, who is head of image analysis facility at the Human Technol in Milan. And Fian, I'm gonna start head of an image analysis facility.

Florian Jug (00:01:45):
Yes, that's correct. That's half my job. I'm also running a research group, more basic research in deep learning and machine learning on bio image data. And then there's also a facility ,

Peter O'Toole (00:01:59):
Because I, I think that's extraordinary to actually have an image analysis facility. I, I, if people have listened to this podcast in the past, you'd have heard how image analysis is the next big thing. And now we actually have Florian who's actually has a facility for image analysis, which I think is really, uh, forward looking, I would say. How,

Florian Jug (00:02:20):
How did you Yeah, that's a good, that that's a good thing when you, when a new institute starts, uh, that you can kind of, um, break with habits and do something new. And I was also very lucky that my ideas were received very positively. And I, I pitched it like this, it is nice to do basic science on new methods, but then bringing them to people is really hard. And in a facility you can, you can directly work with people that have problems, which is also good and informative for the basic research side of things. And then in order to make this all happen, we have actually a middle guild, which are research software engineers, and they are very little spoken about and very hard to find and extremely important to have. And so we, we have a number of people that know how to build research, usable open source software, and kind of they take what we do, but also with other groups do and, and give it as useful tools to our users.

Peter O'Toole (00:03:19):
I, so I I I, I think it's a brilliant idea. Uh, I have a question. Who wants to work in an image analysis facility? What, what type of individual is it that you are having in there?

Florian Jug (00:03:35):
Um, I think there is, in almost every live facility is at least one or two people that are almost working in an image arts facility, just that it's not officially instantiated, but there are always like some Fiji Gus or some other kind of processing inclined people that help the customers also to kinda deal with yeah. All kinds of data analysis issues. And we just, yeah, instantiated it on paper and we work very tight with the light imaging facility at Human. But yeah, I think the more complicated data gets the more important. It's that you have more dedicated time to really think about what is a good pipeline for a given problem. And so we have this time now.

Peter O'Toole (00:04:28):
I, I, I think it's brilliant. I, and the reason I guess I was asking, you've got your research group, which I guess is researching into, well, you tell us. What is your research group researching?

Florian Jug (00:04:41):
Um, okay. At the moment we do a lot of, um, de and image transformation methods with machine learning. We do a bit of segmentation, we do a bit of tracking and, and that is the image restoration part is successful enough that it occupies a lot of our minds, but I would not say that this is what I'm doing, so I might very well, uh, uh, go into different directions. Also, the human technical, um, it's not the most microscopy heavy community. There's lots of genetics and genomics and, uh, structural biology. So that is microscopy heavy, but but much higher risk with a lot of electro microscopy. And so I'm very keen also in growing into this new community and maybe opening new, new, uh, research directions that are maybe even not image based, but more sequence based.

Peter O'Toole (00:05:42):
Yeah, no, I, I think that makes a lot of sense. I think data's actually start the data from these different technology aspects like genomics, metabol, omics imaging are increasingly coming together anyway. So I think the analysis side of is going to become, uh, really important. Mm-hmm. and, and not easy, really conceptu No, it very easy, but practically really difficult.

Florian Jug (00:06:09):
There's a lot of, it's very easy to speak about it, and it's very easy to see that it will be a thing and in quite the thing yet, right. So multimodal data analysis, it's really hard to, how to combine different modalities, mainly if it's imaging modalities and non modalities. And, but I think it, the time will come and deep learning will, in general machine learning and like statistical analysis will play a very important role. And I think, yeah. I, I can't wait.

Peter O'Toole (00:06:40):
So, and you a, a relatively new, relatively young PI of a for a research group, and you are, you are in a relatively new institute as well. Yes. So have, how are you finding that? Well, you know, what, what excites you? What brought you to the human technical?

Florian Jug (00:07:02):
Oh, many, many things. Um, first of all, I'm in the relationship with another scientist, so we have to deal with all the kind of real world problems of, uh, waking up in the same, uh, room and the same time also finding a job that gives us the opportunity to do science on a really, you know, like professional, uh, international level. Um, so we were lucky enough that, uh, so Gaia is very successful on her own. And, uh, we were lucky to find a few things that also, uh, used quite a lot. So I think we made a great package and, uh, the human technical was looking for somebody taking care of the em side of things. And so Gaia was really good there. And, uh, after I pitched my ideas, it turned out that they were also really interested in giving me a fantastic package, actually. Unbelievable. I would not have to write grants and could have a group of 10 people and Wow. Yeah. And, and that's opportunity that it's, it's really good. My wife is also Italian, so going back to Italy and was a, was a positive as well for us.

Peter O'Toole (00:08:15):
Ah, no, I didn't know your wife was Italian. And a group of a group of 10 is, is obviously a very big, uh, big,

Florian Jug (00:08:23):
Almost too large my is to stick to group size up to 10 and or smaller at least.

Peter O'Toole (00:08:32):
I presume that means you can start really exploring very different areas of image analysis or data analysis, not just image analysis, obviously. So I guess that's a good thing. And it's a group mature. You have that support under you. I mean, it's

Florian Jug (00:08:48):
Not a, it's not a research group of 10, it's, it's 10 over the three different sections that should not feel like three sections. We'd like to feel like a group of friends doing things together, but there is like basic science with PhD students and then there research engineering the facility. So with it means like three, three and three plus. And then each individual segment is, you know, not overly, overly large, but still we can explore a lot of things cause the PhD students don't have to worry about like, coding things that people can use. Cause there's people that do that. And, and the facility can really concentrate on taking the tools and also requesting the creation of useful modules that they can then use and bring to our customers.

Peter O'Toole (00:09:41):
I I like the way you demarcate into three distinct areas. And, and I know from my side the importance of that, but you also perceive that as being quite important that they have their, their, their badges that they wear in the technical

Florian Jug (00:09:59):
Yeah, the badges is like the feeling of responsibility and ownership of what and purpose. Right. Um, I really do not care if they sit segregated or, you know, it should really feel like a, a group of people that have their purpose and, uh, together capable of doing more than they would be without the others.

Peter O'Toole (00:10:22):
Right. I, I'd be guessing that if, uh, if someone wanted some image analysis support mm-hmm. , they know who they're going to and they have a, an expectation, whereas if there isn't a solution, they go to the other group and they're at help. And it's a different end point expectation, different delivery time expectation maybe.

Florian Jug (00:10:41):
Yes. It's also a different, um, it's also a really different type of job. So in, in the research group, the idea is to pursue an academic career. So you wanna publish, well, you wanna do something that is also, uh, well perceived by the machine learning or computer vision community, which is not, you have to jump through some hoops sometimes to publish papers. You have to also apply it to some real world data on faces or on on street scenes to be reviewed favorably. And, uh, and the research engineers, uh, sometimes people that have a PhD but don't want to be a PI have incredible skill. And so they also look for a more stable job situation. So Im actually capable of offering long term positions similar. Also, also the really speak with our, with our biologists that, that receive new projects, um, are also typically, um, not at the age of a PhD student anymore. And, and, and, and they seek some kind of amount of ability to settle into a, a longer term life situation.

Peter O'Toole (00:12:04):
I think that's a really good point. I, I've met resistance, uh, myself thinking about similar posts, uh, sort of around the data science side that why would you stay in academia for that salary when you can go, because of the skillsets they have. They can easily go to industry, to their commercial markets, the money markets, banking markets, and make a lot of, a lot more money than they can in science. But, and yet it's, it, my argument there is academics stay in the academic realm, but they could go off similarly into industry and make more money. But it's, I think it's a freedom the, the way that you have some not sweat, you have some influence over where your work goes, even within the facility itself, you have some influence of how you bring users in and you're doing primary, they're still helping primary research, they're still helping solve your cancers, your stem cell research, your ecological research. They're still having all those impacts, but with a Absolutely.

Florian Jug (00:13:09):
Yeah. This is really the, the ideal, the ideal candidate really, really, um, ties into the mission of supporting the life sciences to do a better job, to be, uh, uh, creating methods and tools and creating interactions and collaborations that help to make the most outta data. And, uh, it's really hard, mainly in the research arm of things where our, I would like to hire postdocs that could very well go to Google, Amazon, or Facebook. And, and these jobs are not boring. There is very interesting problems to be solved. And the starting, um, salary is most likely an integral multiple of what I can offer. And so I have zero postdocs at the moment in the research side of things, and I would, if, if a skilled postdoc that has a mission that says like, I would like to solve this problem in the life sciences would come and pitch this idea, if I don't think it's a flaw idea, I would give 100% freedom. And still, it's very hard to find people because there's so, so much need for, for good machine learner, for good people, learner for this kinda skillset at the moment. It's such a hype topic.

Peter O'Toole (00:14:33):
I I I was gonna say be patient cuz you know they exist. Well you exist and you've come through that route, you know. So I'm gonna take you back actually. Okay. I'll just get you back to childhood times. Uhhuh , when you were young, 10 12, what did you want to be career wise?

Florian Jug (00:14:54):
Um, I was not decided, but I knew that computer science is absolutely amazing. I was already coding at home, we were moving quite a lot. Um, so there were always years. I didn't have a huge amount of friends yet, and that was great times to be with my computer. , it sounds so sad, but it was really good time. I didn't regret it. Um, and then it was with times where I did lots of, and where I went out a lot and I was not a few nerd, sorry. Yeah. Plane. Um, um, so that Ill end up, what I do now was not planned very from very early.

Peter O'Toole (00:15:44):
So, so when did you realize, so you into your coding? Uh, my, my son's a computer scientist with maths and I, I, who knows where he'll end up. Uh, he's, he's getting there now. I think he knows where he wants to go. But your fir what was your degree in? Is that computer science?

Florian Jug (00:16:02):
I, I started computer science and I have a minor in logic and philosophy of science

Peter O'Toole (00:16:09):
And I was in Munich, is that correct?

Florian Jug (00:16:12):

Peter O'Toole (00:16:12):
And then, then you went onto, ah, I've gotta get this right now. Is it informatics and management? I've done my research and

Florian Jug (00:16:22):
No, I, uh, from Munich, I went, um, to Siri and I was joining a group that does a lot of discreet mathematics, um, and extremely properties of random grass. Don't ask what it's, I mean I could explain it, but you wouldn't wanna know. But in this group I started doing computational neuroscience in collaboration with, and that was, that was exactly what I wanted to do. And I was kinda the, the scientific hobby of my, of my doctor, mother of my supervisor. Um, but with collaborators at Institu Mathematics, I had really, really, um, good partners to learn and think a lot about how newer networks, how small, including devices that by themselves are very limited in the capabilities in the network can become, uh, so capable. And I, I loved every day of my more than five year PhD, but after five years I looked back and asked myself how much did I really learned about our brains? And it was so little that I expected that I would would've to be like 700 years to make a real dent . And then I, I switched

Peter O'Toole (00:17:35):
Aren't they are complicated beasts, aren't they as brains? You've been in Germany, Switzerland, now in Italy, but I also know you so well actually at the moment. You are where God, where are you at the moment?

Florian Jug (00:17:52):
Oh, at the moment I'm in, in Woodall, uh, at the mpl because we, it's the, actually the last day of a two week deep learning course we're teaching former cross, um, to yeah, to help them kind of understand this new technology and how to maybe use them on their own in the future to help their own analysis.

Peter O'Toole (00:18:12):
Yeah. And and you have a t-shirt to go with it, so,

Florian Jug (00:18:15):
Oh yeah,

Peter O'Toole (00:18:17):

Florian Jug (00:18:17):
T-shirt is actually, you can see here's, and then on the back,

Peter O'Toole (00:18:26):

Florian Jug (00:18:27):
We have to back focal plane and um, that is not from this course, this is from a microscopy course where I am, uh, in the lack position to also teach every spring in Core Spring Harbor. And we acquired this on a, on a scope that we built with the students.

Peter O'Toole (00:18:44):
Wow. So wait, so wait, you mean, so you are, you are in Woods Hole now. You go to Cold Spring Harbor. I know you do the E em b l course. Well, cuz uh, Laura Wiggins, my PhD students who, uh, with me came to what met you right at the start of her PhD. It's a lot. Why do you do so much teaching?

Florian Jug (00:19:03):
Uh, because I'm not in a university, so I don't have to do any teaching. And if you're not forced to do something, then maybe it becomes more attractive. I'm not sure. Now I actually, I do a lot of teaching because it's really a lot of fun and, uh, the whole endeavor is built on use, building useful things for people and how would I know what is useful for people if I would not speak with people. And this is a good way to, to get this to understand what are open problems, what do people starting with. But at the same time it's, it's also just a huge amount of fun to, to teach in places where really motivated students go. Where you have, you know, this like can-do spirit, the unbroken young love for science. It's amazing. I love it.

Peter O'Toole (00:19:54):
Yeah, I, I, I think, think you're right actually, if you think about undergraduate lecturing, uh, yeah, they choose a course, but they don't necessarily think that each module is perfect for them. So they're not all a hundred percent engaged, whereas the courses you are running, people are volunteering or even paying to go on because they want to learn that specialty. So, so it is, and and ultimately, I, I know from Laura when she came back, they're inspirational for the students. Uh, it enables them to step up. So, so watch out for her work, uh, also in segmentation and tracking. Uh, but something different is missing at the moment. So looking

Florian Jug (00:20:31):
Forward, I was a student actually at, in for the physiology course, which is a seven week course. And you start at nine in the morning and, you know, you have down times, it's, it's, it's like seven weeks is a long time and you get to know other people really well and, um, it, it was, yeah, it was transformative. Uh, course courses in, in like Course Harbor or here in Wood Hole, uh, yeah, you, you cannot be the same after the course. You'll be different.

Peter O'Toole (00:21:01):
No, no. So actually I did an, uh, E embl course, EMBO course at Eem b l back in 2001. And that was very much career inspiring with Timmo Zimmerman. And actually I think Ricardo Henrique was at that call Yeah. As well as a student. Uh, so I, I don't think I, he was in different, a different group, but I think we were at the same course together, which we hadn't realized at the time.

Florian Jug (00:21:23):
Yeah. Ricardo ran actually an embo course next month in Portugal. And I'll also be be there for,

Peter O'Toole (00:21:31):
Right. It, it's, uh, yeah, they're, they're, they're just brilliant and thank you for actually running them. So there we have it. That's where you started, that's where you got to. Where do you see yourself heading?

Florian Jug (00:21:45):
Um, okay. I'm the first time in a situation where I actually could stay physically in Milano for many years to come if that will happen or not, God knows. Um, the, um, facility plus basic research thing is, is a bit of an experiment. There's very few, I don't know if anybody else that has a similar setup. Um, I hope that this, uh, will prove to be extremely productive and fruitful. And then, and then that might be something that is enjoyable quite a long time or I grow out of it and would like to maybe make it help other people to do similar things and, and help help the community to more this model more often in different places. I'm not sure.

Peter O'Toole (00:22:39):
So you never attempted into industry?

Florian Jug (00:22:43):
Oh, it was attempting many times, but then I think I got protected from jumping into a better paid job with the opportunity to actually have weekends by being married to an biologist.

Peter O'Toole (00:22:59):
Uh, you're telling me you get weekends?

Florian Jug (00:23:04):
No, I would have gotten weekends if I would've jumped into industry, in the industry job. Right. But, so I do have weekends, it's just, there's always more to do. Right. And I have a bit of an addictive personality and I really like what I'm doing and I like to do it well. And so science can be a very time fulfilling hobby.

Peter O'Toole (00:23:30):
So yeah, good term to use. Uh, a a job fulfilling hobby. Uh, you know, so it shouldn't feel like work, you know? Uh, sometimes it does cuz there's always the this stuff you have to do alongside it. So it's good to hear, you know, yes, you might do it at the weekend, it might intru, it's not intruding cause it's what you like to do. So thinking of hobbies, what other hobbies do you have besides science?

Florian Jug (00:23:56):
Uh, I run a lot. I like running a lot. What distance, uh, what distance would you like me to run?

Peter O'Toole (00:24:06):
Go, go on. Let What's your typical training week and what's, what's your biggest event?

Florian Jug (00:24:13):
Um, it really depends. At the moment I run maybe between 50 and kilometers a week. Then there is times where I run less, but there certainly often times where I run much more. And my goals are anywhere between a fast 10 K, which is actually really painful to train for, or a very slow and relaxed 24 hour race. Oh, hundred hundred k race.

Peter O'Toole (00:24:41):
Okay, so, so I, I I see now, now you're at I'm very similar. . My latest, oh yeah, my latest medal just there is, uh, is a 24 hour run.

Florian Jug (00:24:53):
Amazing. Yeah, I did only one. I was actually really a lot of fun. I was in the preparation is, is really interesting because you have to really get to know yourself and ask yourself, what might I crave? What might I be able to digest and want to eat and not vomit,

Peter O'Toole (00:25:12):
Or have other problems, which, which I encountered actually. How far did you cover in the 24 hours?

Florian Jug (00:25:18):
Uh, too little to be proud of. Hundred 40 something, hundred 41 or so kilometers.

Peter O'Toole (00:25:25):
Maybe, maybe you should come over and we should do this one, which is just down the road not far from here. Uh, our target was 100 miles.

Florian Jug (00:25:34):
Yeah, I, of course I also wanted to do a hundred miles, but I, I kicked the stone really hard in the first marathon. And then the second marathon, I thought it's, I just kicked the stone, but I think I, I, I compensated for the ankle that hurt a bit and then just everything fell apart. It was very painful, . And then in the, in the night it started raining and everything, everybody went to bed. And then, and then I, I was not strong enough to stay awake, so I went into bed for like two and a half hours. No. Oh no. Yeah, it's, it's a shame. But I, you know, the hip started hurting, the knee started hurting and, and everybody was bed and then there was the rain. And in one loop, I, I didn't take the right turn that I took like 40 times before and I went straight on for half a mile and I was like, I go to bed, Gloria. But then I regretted it very much, you

Peter O'Toole (00:26:28):
Have to do another because

Florian Jug (00:26:29):
I have to do another. It's true

Peter O'Toole (00:26:31):
For our 24, we, we never, we didn't stop. We, we, we, we grabbed some food and drink, but it was lap after lap. And I don't think once, uh, between me and my running partner, we running together. So it's solo. Not once did it cross our minds, we were gonna stop. And my friend unfortunately got really injured very in the end. And wow. He pained it through for the last lap and a half. We, which lap and half is still seven and a half. So it's, it's still 1213 k and it was slow going, but we got there, we, and we ticked a hundred miles within the 24 hours. It was, uh,

Florian Jug (00:27:09):
Congratulations. Yeah, I'm a bit, yeah, it's nagging that I didn't do the hundred miles and I certainly, and I think I wasn't in shape to do it. Um,

Peter O'Toole (00:27:20):
Next year. Yeah. If he'll kill me if I did it, but try .

Florian Jug (00:27:28):
Yeah. I'm not sure how guy would react. I think it's okay. I think she understands.

Peter O'Toole (00:27:33):
Uh, yeah,

Florian Jug (00:27:34):
It's a lot of time. It's a lot of time. I also trained this this spring for really nice race in, in in in Tuscany 105, but, but 10,000 feet of climbing and I got Corona two weeks before it.

Peter O'Toole (00:27:51):
Ah, so you didn't run it horribly?

Florian Jug (00:27:53):
You didn't run it. I couldn't,

Peter O'Toole (00:27:55):
Yeah. Cause cuz it does take it. Yeah, it's not, I, I ran on Cor Road a bit short distances only,

Florian Jug (00:28:02):
Uh, yeah, I, I I was lying in bed for like 10 days and it was two days after and it's not the best tapering to not do anything. Yeah. I didn't do it. I was also anxious. I, I didn't know how I would react and, and then I, I would be too headstrong to stop even if I would feel something and I, I was not sure if it's a good idea so I didn't do it. But I'm in great shape now,

Peter O'Toole (00:28:24):
, which, which is good. It's only consolation. I was meant to do the four hour run last year, but I broke my ankle the Monday before it. Oh

Florian Jug (00:28:32):
My god.

Peter O'Toole (00:28:34):
So all that training and it's insane training, isn't it? It's, it's so much discipline all out the window like that. But we were back. Anyway, moving on. What other hobbies do you have?

Florian Jug (00:28:46):
Uh, I, hobbies for us really come moment it's a pottery high and then there's photography, which is at the moment, at the almost all time low.

Peter O'Toole (00:29:00):
So you have a picture I think of your pottery.

Florian Jug (00:29:04):
Oh yeah, I do.

Peter O'Toole (00:29:05):
Um, blast out on your, your, your background for us and tell it, because surely this is your photography and your pottery together. If you took the picture.

Florian Jug (00:29:13):
Yeah, I took the picture. But yeah, with an iPhone. But I, I do like taking, taking pictures and it's a really beautiful way to kinda share your way of seeing things. So lemme see some background and then here we go. Ah.

Peter O'Toole (00:29:31):
Oh, do you know, it really does, honestly, I can now see it enlarge. It looks like an arm bread

Florian Jug (00:29:37):
. Yes. It, it's brilliant. I this this doing these things because you, you want to control the shape of the object very precisely. And then you it and you put it in the wood firing where you have very control of how the texture would evolve and how get out. So you kind of combine your unreasonable desire to control the world around you with the willingness to give your precious peace into a very uncontrollable, uh, uh, final process. Process. And we fired the killing ourself, which was the first. So we were like, uh, it's a five day process. There's like more than two days of, of stocking the kiln, which is the oven. Um, this is also why people say I'm stoked because you're kind of fired up and learned. And uh, it's 1300 degrees Celsius, which is a huge, uh, temperature to reach only with burning wood. So yeah, it's amazing.

Peter O'Toole (00:30:46):
And is that a, do you have your own kiln if you state it yourself or is it somewhere else you did it?

Florian Jug (00:30:52):
We do have a kiln, but it's an electrical that we can, it's more electrical at home and this, uh, wood firing cleaner. Yeah, they're beast. I mean you need a big garden, but I I, we have this plan. We are currently in the process of kind of buying a house and it would come with enough garden that we could probably put a kil in there and disguise it as a pizza oven, which is much more tolerated in Italy than having a garden.

Peter O'Toole (00:31:23):
You'd probably beat to me too. It, cause I was thinking if I had a kiln like that when I finished with it, I'm sorry, I'm throwing my pizza in it. Cuz that would be what, one minute come start to end and that would, yeah,

Florian Jug (00:31:34):
I guess it would be a minute from start to not existing anymore. It's, it's brutal. It's so hard. Degrees is brutal. You open the, the hall where you throw the, the wood in and it's you like a meter 50 away and it's like the devil is licking your face. It's, it's so strong. You have no notion of it. It's incredible.

Peter O'Toole (00:31:58):
Yeah. So how long have you been doing pottery for?

Florian Jug (00:32:03):
I started doing my PhD in sew before I went to Western.

Peter O'Toole (00:32:08):
So it's a long term hobby then. It's not, you said it goes in waves and cycl, but this is quite a long cycle.

Florian Jug (00:32:15):
Yeah, 12 years we are in maybe cycle four .

Peter O'Toole (00:32:20):
So does that mean you've gone outta pottery and back into pottery over that time?

Florian Jug (00:32:24):
Yeah, yeah, yeah. Absolutely.

Peter O'Toole (00:32:25):
What got you into it?

Florian Jug (00:32:28):
Um, there was a sign next to the street that said pottery lessons and Gaia said pottery lessons. And then I thought, oh, she responds to pottery lessons that might be a good present for her. And so I gave her, yeah, we went to this pottery lessons for a long time. Actually it was, if I say pottery lessons, you imagine 20 people following a teacher's voice. It was a really, uh, really capable potter and she would open her studio for a bunch of scientists and it was, it was delightful to be there with like, there was professor for neuroscience and postdocs and a few PhD students. We were usually like maybe seven people and, and the artist was an amazing personality and very capable. But the thing she was certainly not was a teacher. So we, we would just do what we wanted and only when we did something that, that she found offensive, I guess she would say like, why can't you try to do this? Was very, it was very nice.

Peter O'Toole (00:33:33):
So it's good networking as well. Yeah. So again, very useful. I'm gonna bring you back into science for a bit and we'll come back out in a moment. So obviously you are into Denoising images, uh, you've done quite a lot into Fiji and different plugins. I usually ask a guest what's their favorite publication. I'm gonna ask you in your case, what is your favorite bit of software that you've let Yeah. Opened up to the community? Do you have a favorite?

Florian Jug (00:34:06):
Yes, I think so. Um, I think , um, one of the de tools, I really like it. I think the idea is very elegant and very simple. Um, and it's extremely useful. I I really get a lot of kick outta doing things that beautiful from computer science and computational standpoint, but it's really a different level if it is also useful and people start using it. And I think most of all it's useful quite a lot.

Peter O'Toole (00:34:39):
And how long did it take to develop that?

Florian Jug (00:34:44):
It was not so dramatic. Um, I mean, we still develop it in some sense. It, it had a whole like slur of like, you know, versions and, but from the, the idea was brought to me by a postdoc at the time, Alexander Kool. Um, he, he looked at the care work which came before that, the image restoration. And he was like, but we don't need ground. We can do that. We can just take a body of noisy data and and figure out how to, how to de-noise it. If the noise is pixel independent with noise and noise and noise and noise is, and from the moment where I said, nah, are you sure? To the moment when we published was maybe five months.

Peter O'Toole (00:35:31):
Wow. That's really tough. And

Florian Jug (00:35:33):

Peter O'Toole (00:35:34):
So okay. So if that's five for your favorite bit from start to end, you sort of mentioned sometimes it's the simple ideas. What's the longest project you've worked on that you thought would be really good, but actually maybe, maybe you haven't actually got to the end yet. Uh, uh, to a point that you can actually publish it in some way.

Florian Jug (00:35:53):
Uh, my PhD thesis certainly, but that is a different topic. Uh, and in the, in the bio image analysis arena in as a postdoc I worked a lot on track and uh, I know that the field and the problems very well. And I think we have wonderful solutions worked out and then we never, and then care happened. So Martin Vigar at the time had this wonderful idea of using units to do image restoration and, and it was immediately clear that this is so obvious somebody will do it soon, but why not doing it quick ourself? And it lifted off much crazier than we ever expected. Um, and from that time on, image restoration was a huge, um, time sink I wanted to say. But of course it, it, it was a beautiful thing as well, right? But, but a lot of, of the group, the young group, it was the idea was born just before I opened the group. So I was still doing it in Gene Meyers's lab mm-hmm. together with Martin and u schmid, thousand other people that gave us data. Um, and that kind of pushed the tracking away. And only now I have a first student again and you wanna look into, into tracking a bit more and I'm actually excited that it comes back. But, but it's years, right? Five years or something.

Peter O'Toole (00:37:23):
I, I don't think it's just fascinating. I think, uh, if any, if any PhDs or lay audience are listening, that sometimes some you have a great idea, but it can be really hard work to get to the end. And thens accounts can come in so simply and get to get to fruition really fast. It, it's just silence is, it's just an almost, you can't predict how long something can take in insides. It's really complicated.

Florian Jug (00:37:48):
You notice machines when you, when you, where you throw quarters or whatever country you're in, coins in and they fall on this like, uh, platform and there's this small thing that kinda shifts them in front and then they're like at the, at the edge at they're like, it cannot be more than one coin until it falls. Right? It's like it's there, come on, give one coin. Gimme a coin. And you try and they fit more and more coins and it just doesn't follow. And I think science is a bit like that cause noise. What would not have happened in, in this short period of time if there was, if it was not built on top of a lot of thought and thinking and coins that did not fall off the edge. Right.

Peter O'Toole (00:38:24):
I love that analogy because as you say, sometimes you put your coin in and nothing comes out but you keep going Cause you can see the big prize and sometimes you put coins in and you'll get one coin back or maybe you'll get a few coins back and you think, no, I'm not stopping yet. I'm gonna keep putting Cause I can see the big prize that I think, and that's a really useful analogy cause there comes a point if you keep waiting for the big prize that sometimes never comes, that data never gets shared. That work, that research never gets shared. So there's a point where you have to publish, uh, release the work. Uh, and as you say, what you've done with uh, some of the de-noise and stuff is you then keep on iterating and polishing it. You don't wait for the, the perfect solution. Cuz ultimately the perfect, I guess in your case especially the perfect solution is honed by the people who pick up your software and use it and feedback.

Florian Jug (00:39:16):
Yep. I fully agree. Science is such a weird thing, right? It's, it's kind of, it's a search algorithm. We search for things that were unknown before and unknown now or in, in, in, in my, it's also finding things that enable looking at data in different ways or kinda make, make, make, make your life easier to discover things hidden in microscopic data or hidden in data in general. And it's weirdly implemented on, on a social community and it's, it pays off to be persistent and bang your head against the problem even if it resists at first. But there's also problems that will resist forever. And I think it's a really delicate balance. Do you go on or or do you kind of realize that it'll not happen? And if you, if you do it too fast, then you'll never penetrate very deep. Cause you'll kind of try, try, try and, and usually it's not, it's never super easy, right? So you need to persist it. We would call it branch and bound computer science, right? You branch into different ideas and then you bound the ones that are hopeless. But what is hopeless, right? The the solution might be just one side away. So that's

Peter O'Toole (00:40:36):
Interesting. And then there's the importance of having multiple branches and, and not just, I think so one route because as you say, some things don't work. So one of the reasons that so much of your work has been so widely adopted is a lot of it is freeware. You know, you, you put it out open, uh, open access to what you're developing. Why did you choose that route over binding it up and selling it onto one of the big companies as a a, a commercial product?

Florian Jug (00:41:12):
Um, it's, uh, that's a really good question. Okay. There's so many thoughts. How do I linear linearize that out in one three more thoughts. It's the only way to, to be really reactive. Uh, open source software can be messy and disgusting, but at least it's adapting fast to, to whatever needs to be done, right? Fiji is the most wonderful, shitty piece of software is what I to say. There is a thing for everything. And if, if, if you would a new tool or a new data modality, if a new microscope comes along, it will be a plug that does something meaningful with it. Relatively soon somebody will do it. Ooh, do you hear the tower?

Peter O'Toole (00:41:58):
Only? Only vaguely or the,

Florian Jug (00:41:59):
Okay. Okay.

Peter O'Toole (00:42:00):
So yeah, your, your headphones are also denoising your background

Florian Jug (00:42:04):
. Okay. That's very good. Yeah. Uh, so, um, you want to react fast to what is needed in science because science by definition should be dynamic. Um, but, and the people that contribute are, you know, they don't know each other. They have a different coding style. And so everything becomes a patchwork of of it's not clean, right? And if you accumulate in Fiji is not the youngest or softwares or image J um, if you accumulate this many things, you have something wonderful but also wonderfully diverse. And while diversity can be beautiful in software development terms, it's really, really old . It makes everything hard. And, and then also funding, it's not common today to get money from our funding agencies to keep software alive. And in Germany we slowly start, and I have no idea why its av I guess because I was there for the last 10 years.

Um, the, the dfg now starts funding software maintenance. But that's really, that's really, really uncommon. And I think we have to do that more in the future because we, we, science is not anymore like, you know, you look through a achieve lens and, and, and observe something not new. You have to dig deeper. And if you dig deeper, everything becomes a bit more complicated. And I think the data analysis, we need some tools that kinda go with us and grow with us. And they cannot grow with us if you cannot maintain the basis of them. And our assets change and, and you know, there's new versions, uh, of, of, of windows and que and, and everything falls apart cause it's this like bubble that we create as humans. It depends on each other and it's, it's disgusting in many ways. And so it, without this funding, it'll be very hard to keep digging deeper and deeper efficient.

Peter O'Toole (00:44:03):
I, I don't wanna get too technical, but yes, , if if, if you could start Image j Fiji today, would you use the same coding language?

Florian Jug (00:44:16):
Ah, you put your thing on a really delicate spot. Um, so Image J and Fiji and also other tours are based on Java and that was a great choice at the time. And now the whole deep learning world, um, embraces Python very much. And Python is amazing for many, many reasons and disgusting for others. But it is the new thing and it's very hard to find people that even are good Java programmers today. And it's, everybody has some experience with, so this is exactly what it means. See now the world changes for some reason or another and a lot of things we did, uh, falling apart and, and kinda out date slowly,

Peter O'Toole (00:45:01):
Which is kind of why I was asking the question of, yeah, it it's how do we migrate from, from using to Java based into Python based cuz so much will you've, you said so much exists in this, this this sphere of image j Fiji under based under Java. But a lot of the new scientists, a lot of the new scripts coming are gonna be coming up under Python and they're not gonna be that compatible. So yeah. Do you think a second Fiji

Florian Jug (00:45:32):

Peter O'Toole (00:45:33):
Fiji p Proprie will come through and they'll coexist and then slowly migrate as people patch things across or duplicate things across,

Florian Jug (00:45:43):
Essentially this is what has to happen. Yeah, we have to redo a lot of things and it sounds horrible, but on the other hand, um, the scientific community also works over, right? You get older, you die, young people come along and, and even if it would not have to roll over to a completely different program language and new people that come in have to relearn everything that exists. And so relearning a lot of existing things is also not as much fun as creating something. So maybe it's not as horrible that that, that things roll over and and renew themselves. And also we have some lessons learned. Why is Fiji not only the most wonderful piece of software but also horrible? Maybe we can kind of avoid some of these mistakes and starting from scratch is not always bad is what I want to say. It it, but, but the problem is that it requires a lot of, a lot of time and a lot of dedication and so we need to give people that need to do this job, the opportunity to be happy in their jobs.

Peter O'Toole (00:46:46):
Yeah. Cuz cuz you know, ImageJ Fiji isn't just for mic, it was built, microscopy was very much a foundation, foundation to that. But all sorts in the world of science use it for all sorts of things. From looking at their gel scans to their photography hobbyists. It, it's now used by so many different variety of people across the, across the sciences and hobby, well it's so

Florian Jug (00:47:08):
And so does Python, right?

Yes. Python is used in many disciplines as well and, and uh, nap is one of the very well funded developments in the recent years that I think has a good chance of, um, becoming indispensable for many of us. Is it replacing Fiji? One has to see at the moment. It's not this like battery included many plugins, but I think some people in the n space, um, see the future, um, in that very close to kind of being a feature replacement in the, you have to see it's a bit early, uh, to, to judge Yeah. If it'll be successful, but I think it has good there has good potential.

Peter O'Toole (00:47:55):
So it's the evolution, but it's a big evolutionary step in this case. Uh, which which isn't always common. I'm gonna ask you some quick fire questions, uh mm-hmm. , so I hope you're ready for this. Are you, uh, actually, do you have any bad habits besides running for far too long?

Florian Jug (00:48:14):
Oh, I'm sure, but I'm really the wrong person to, to be asked do bad habits that I know about. I quit smoking a long time ago. Yeah. I think the running is maybe the worst of it.

Peter O'Toole (00:48:29):

Florian Jug (00:48:31):
And then what could say that not, not being able to, to have a good worklife balance is maybe a better habit. I'm not sure I have to work on that too because I'm getting older. I noticed that I, I have to think it bit easier. Maybe

Peter O'Toole (00:48:44):
I Yeah, but you said that your work is also your hobby. So I think the balance,

Florian Jug (00:48:50):
It's true the last two years were very stressful though. I think it's also cause of the move. A lot of things change and, and you cannot only like run into science, but you also have to take care of new insurances and all this like, annoying things that come with being a human in our society. And maybe that's stressed me out more. I think when this force away then maybe I can settle again into a formula status between long hours in the lab and running.

Peter O'Toole (00:49:16):
Yeah. And you set up a new facility, a a a new lab and a facility within that lab space and you know, you've moved to country, you've moved to countries, so the bureaucracy will be different and there's a lot of learning when you change institutes, even within the same country. There's so much ways of working. I can imagine that's been a quite a titanic learning curve, uh, so early on as well. So I wouldn't worry about it. It ne that doesn't get easier, Gloria me will not get easier.

Florian Jug (00:49:47):
Absolutely. Challenges. I just, I just focus on the processes. Italy is an amazing country. Um, you can go in places that wherever you go the coffee is like through the roof. Exceptionally amazing. Mainly if you kinda compare to what I had in the past 10 years, um, 20 minutes by car we are in, in not in the, in Switzerland. We can do fantastic hikes an hour and I think it's the potential to balance is phenomenal. , let's see how I do

Peter O'Toole (00:50:23):
. Okay. More quick fire. Are you an early bird or night o

Florian Jug (00:50:30):
I transition in grow older night owl. Absolutely. I would've said until maybe two years ago and now I'm in a transition phase. I think I'll very soon be old enough to stand up early and go to bed late.

Peter O'Toole (00:50:42):
Yeah. Welcome to the world. No, it should be PC or Mac.

Florian Jug (00:50:48):

Peter O'Toole (00:50:49):
Mac, McDonald's or Burger King.

Florian Jug (00:50:52):
Burger King.

Peter O'Toole (00:50:53):

Florian Jug (00:50:56):
Yes. Mainly since they have different fries.

Peter O'Toole (00:50:59):
Oh no, it's a fries. You letting 'em down

Florian Jug (00:51:02):
Cheese. Insane. Enormous. Ah, yeah, no, totally. Absolutely.

Peter O'Toole (00:51:10):
Uh, coffee or tea?

Florian Jug (00:51:13):

Peter O'Toole (00:51:14):
Uh, beer or wine?

Florian Jug (00:51:20):
Don't care. Flat prior, but everybody's available. Okay. Not mixing at all.

Peter O'Toole (00:51:25):
Alcohol? Yes. Is

Florian Jug (00:51:26):
The alcohol? Yes, absolutely. I'm a good friend of, I went to a very good school.

Peter O'Toole (00:51:32):
Ok. Chocolate or cheese?

Florian Jug (00:51:37):

Peter O'Toole (00:51:38):
Okay. If so, woods Hole. There's going to be there. There's ban of being a tutor's dinner where everyone goes out and it'll be a nice place to eat. What have be the best food that's been selected for you to be put in front of you? What would be you think that is just wonderful. What's your favorite dish?

Florian Jug (00:51:55):
In or in No, just in general.

Peter O'Toole (00:51:58):
No matter what, you know what it's like, you're invited to talk, you get taken out and then, oh, and you get no choice of what to eat, just put in front of you. But what would be the perfect dish that they could put in front of you?

Florian Jug (00:52:11):
Um, if you're a, okay, you can also put it in other countries. It's like a a t steak with the file and the counter file on the other side. Ideally four, five centimeter toy and then brought to a very hot stone for like a few seconds on each side so that the, the inside is still essentially, yeah, only like warm, but not, oh my God. I'll start.

Peter O'Toole (00:52:44):

Florian Jug (00:52:44):
That's my favorite. And then olive oil, pepper a bit of, yeah. Salad next to it. Best thing.

Peter O'Toole (00:52:54):
There you go. For anyone who's thinking about inviting Florian to talk, you now know what you have to deliver. What is your nightmare though? What would be the worst thing they could put in front of you?

Florian Jug (00:53:07):
Oh, I would have many things to say as a kid, but now I'm very, I really like to experiment. I don't think I, I don't think you can really fuck up.

Peter O'Toole (00:53:22):
Okay. Nothing that you would, uh oh, nothing.

Florian Jug (00:53:26):
Nothing in this nightmare. I mean, uh, there's like fish that is fatty like sharks sometimes or so it's not the most, you know, but I can, I guess you can enjoy it anyway. Yeah.

Peter O'Toole (00:53:41):
Who cooks at home, you or your wife?

Florian Jug (00:53:45):
Um, very early we noticed that we cannot cook together. We have to have a clear boss and like somebody that cuts and does like no planning in the kitchen. And for a long time we, we did it 50 50 and now I turned out to be more than morning person and I make elaborate breakfasts and Guy is usually more the evening cook.

Peter O'Toole (00:54:09):
Okay. And next quick, five question. Book or tv.

Florian Jug (00:54:15):
I would really like to say book, but it would, it would be a lie. Tv.

Peter O'Toole (00:54:19):
. There's nothing wrong with tv. So what, what do you like to watch on tv? Uh, uh, what, what shouldn't you admit to watching, but you secretly like,

Florian Jug (00:54:28):
Okay. There's, there's two things. I really like documentaries that are well, well made. Um, but then sometimes when you are home you just need to somehow cool your brain down to be able to sleep and then it cannot be stupid enough. Binge watching things is sometimes really fun,

Peter O'Toole (00:54:48):
Right? I, I'm with you on that. And actually I'm not totally with you on documentaries. Do you not find documentaries really slow and repetitive?

Florian Jug (00:54:57):
Oh, it really depends. It really depends. Like, okay, let me find one that is really entertaining. There's search for Sugarman Apartheids in South Africa. Um, there is one, one album that is getting bootlegged and every household has it, but nobody knows who the artist is and there's just like rumors. Then apartheid ended. One of the kids Grew Older is a documentary, uh, filmmaker and, and looks for who this artist is and what is real story, wonderful music, really great like backstory. It makes you very interested to kind look about, look up apart Wikipedia, but it, it's very entertaining at the same time. And it's a real life story, right? It's such a crazy story. A whole country doesn't know their favorite artists that kind of defined the generation. It's just insane. I like it. The reality of it is what gives the appeal.

Peter O'Toole (00:55:55):
So thinking of favorite artists, what's your favorite music?

Florian Jug (00:56:00):
Oh, that's also a really mean question. I sometimes say that I have no music taste because I, I like it, I like so much, but I think it's not true because in every segment there is things I like and I don't like, so I must have some sort of taste. It's just very different. Coding works really well. Works really well with like Metal for example, right? Ah,

Peter O'Toole (00:56:24):
Now who was it who mentioned? Oh, one of the other guests actually mentioned someone who actually it is on computer-driven music and I can't remember who, I'm gonna have to think about who that was. Put it on the, uh, snippets at the end after this.

Florian Jug (00:56:38):
But then sometimes you can also code really well with like, with like very monotonous electronic music, but it has to be like, beat because at least when I code, I get in this flow space where you really kinda, you know, fast and it kinda gives the pace of your and love it. And then for running, I listen a lot of music while running and there like it much more melodic, maybe guitars. Yeah. Uh,

Peter O'Toole (00:57:03):
What's your favorite film?

Florian Jug (00:57:08):
Um, I will have to think a bit longer. You have to cut it out at the end. My, I really like, I really like, um, uh, how was it called? It was about, um, um, a student of music. He's a drummer and there's a teacher that is very emotionally draining and brutal to him. Whiplash. Whiplash.

Peter O'Toole (00:57:35):
Okay, so you like Whiplash? Um, the Phil Whip,

Florian Jug (00:57:39):
I like Whiplash. Whiplash. The film is, is an amazing film because it makes a really complicated topic, the main topic. And that is how brutal can you be as a teacher and still, yeah, I don't know. You have to watch movie. I guess there's a reason why it's 20 minute movie not summarized well in, in 10 seconds, but it's a wonderful, yeah, I like it.

Peter O'Toole (00:58:04):
I'm really disappointed that you didn't say Ghost. Now someone who's into pottery, surely it has to be the most famous pottery scene ever is in ghost.

Florian Jug (00:58:12):
Ghost is okay, but it, but every time I say that I that pottery is a, a hobby. People say, oh, ghost

Peter O'Toole (00:58:20):

Florian Jug (00:58:21):
Everybody knows this scene. And yeah, it's totally not my favorite movie. .

Peter O'Toole (00:58:25):
I I I'm looking forward to the parody that you do some point and forward it onto Twitter yourself and your wife just on a, on your pottery still. What's your favorite Christmas film? Do you have a favorite Christmas, Phil?

Florian Jug (00:58:36):
Um, uh, yeah, it's um, um, diehard. It must be diehard.

Peter O'Toole (00:58:43):
Ah, good choice. Yeah, no, that's perfectly good. Now you travel a lot. We work. So going back to work, we, we are, we are actually up to the, the hour, but I'm gonna just indulge for just two more minutes. You travel all around

Florian Jug (00:58:56):
The world also, you can also cut out, you can also cut out a few of the boring things and keep

Peter O'Toole (00:59:00):
People can fast forward, huh? Uh, you traveled all around the world teaching on courses. You're in Portugal, you're in the US at the moment. Uh, you've obviously don't quite like the European countries. If you could live and work anywhere in the world, where would you take your lab as a location

Florian Jug (00:59:20):
At the moment? I wouldn't to leave Milano. I think there is great potential. It's unkept and it would be premature to even kinda start thinking about going anywhere else. I think it'll be a fantastic place.

Peter O'Toole (00:59:30):
What about you retire to, if you retired, where would you like to retire to? What would be your perfect location?

Florian Jug (00:59:38):
I, I really like, um, um, the area around UNC and North Carolina, uh, Durham Carro. It's a beautiful, it's beautiful to run are also very, then the area has many, it's also extremely beautiful. So have you been running?

Peter O'Toole (01:00:01):
What, what you mean? Uh, where you been out running where? Woods Hole at the moment?

Florian Jug (01:00:07):
Uh, yeah, here it's a bit you have to run. Um, got 10 K to go to, to be in a area, offer trails. So I went only one time to, to this area because you have at least half marathon ahead of you if you wanna even reach and come back. Um, so it's a lot of road running along the sea. It's beautiful and Naps Beach. Wonderful. And before I was actually visiting the Bay Area and yeah, I did. That was just so beautiful. I mean, you cannot, you can only fall in love with the . It's, it's amazing.

Peter O'Toole (01:00:40):
Great. We, we are over the hour, but Florian, I'm going to be really cheeky and ask the microscopy. We're gonna do a little subset of the microscopy in the future. Mm-hmm. , I hope, which will be more subject based. And I know you're doing a lot on the light and the EM restoration image side. Uh, it'd be great if we could get you back at some point to talk more about that work and the field of image analysis in that area. So will you Absolutely. And do that subset on the microscopist.

Florian Jug (01:01:09):

Peter O'Toole (01:01:10):
Fantastic. Florian, thank you so much for joining me today. Everyone who's watched or listened. Uh, I hope you've definitely enjoyed it. I think it's been enlightening to hear what Florian gets up to outside of work as well as work. We'll hear more about his work hopefully in the future. But also don't forget, you can go and listen to some of the people that Florian worked and teaches with, such as Anne Carpenter, Ricardo Enriques, uh, Paul, Roger, Mark Ray actually as well, uh, very much in this field area. Go and have a listen to those podcasts as well, Florian. Thank you.

Florian Jug (01:01:41):
Thank you very much for having me.

Intro/Outro (01:01:44):
Thank you for listening to the Microscopists, a bite-sized bio podcast sponsored by Zeis microscopy. To view all audio and video recordings from this series, please visit bitesize bio.com/the microscopists.

Creators and Guests

Florian Jug (Human Technopole, Milan)