Talking Biotech Podcast 405
Dr. Mike Tarselli, CSO Tetra Sciences

Adjusting ot the Chaning Biotech Landscape
===

Kevin Folta: [00:00:00] Hi everybody, and welcome to this Week's Talking Biotech podcast by Colabra. Over the course of time biology and the way we look at biological processes has really changed, and today the modern technologies are really starting to take a center stage in crafting our hypotheses and helping us develop more focused hypotheses and coming up with new tools to test those hypotheses.

There's so many interesting things that are happening in the field of biology, and I thought I would like to dedicate a, an episode. To talking to somebody who's been in many different facets, who's participated in many different aspects of, of how biology is currently being looked at, drug discovery, all the good parts of this, and have a conversation about where things are and where things are going.

So we're speaking with Dr. Mike Tarselli. He's the CSO of Tetras Science. Welcome to the podcast, Mike. Hey, Kevin. How you doing? Yeah, doing really well. Yeah, this is really great. I, I just, for for what it's worth, I, this is the first podcast I'm [00:01:00] co-hosting with my seven week old daughter. So she's here, she's pretty quiet, but if you hear her then we'll just keep rolling.

That's, 

Mike Tarselli: that's lovely. Congratulations to you. And, hey, while I have a chance up at the top of the episode, congratulations on passing the magic old number 400 for podcasts. That's, that's a long slog for anybody. And to do it with grace and dedication is, is a testament. Well, thank you. 

Kevin Folta: That's very nice. I mean, if you, these days, if you make it past 12, you beat, you've beat the average, right?

You've beat the field. Well, let's start out by talking about your current position as C S O. What does Tetras Science do and can you tell me a little bit about your role 

Mike Tarselli: in the company? Yeah, Kevin, thanks. Now upfront, I don't want this to be an infomercial for the company. I'm just gonna tell you some about my role and how it integrates and then we can have a longer discussion about how that sort of helps with the field.

But generally speaking here at TetraScience we are a scientific data cloud. So what does that mean? We think that a lot of people are still stuck in the Excel and PDF and text file and GraphPad Prism world of [00:02:00] biology and chemistry, and that maybe they'd like to break out of that. And, you know, as you know, when you are the person at the bench generating the result, the sort of 1995 version of that was, well, I'll generate it for my assay, my result, my endpoint, and then I'll tell some other department what I've done, you know, usually by PowerPoint or on a acetate slide deck or something.

And then eventually I'll, I'll, you know, move that along through the process and put it in a database somewhere, and hopefully someone will dig it up and use it to file a, a new drug application or N D a. We think that that has sort of been upended and that everybody has a right to all the data that is generated inside of a pharma company.

So in this context, right, you want to move everything to cloud first. Preferentially, like if it comes off an instrument, scoot it up to the cloud, you can still get at it to analyze it, to look it over, to think on it, to, you know, do some downstream processing, et cetera. But, but you should move it somewhere where everyone can get at it.

And then you should be able to surface [00:03:00] all the metadata about it. So sure. You ran that plate reader on this array on a 96 wheel plate or something. Well, maybe you wanna say, you know, Kevin ran that at this time, this date, with this assay in mind, with this dilution curve, with, you know, these samples and this array.

And you started it in this lab notebook and you know you have it for this project, which is against this target. You know, that kind of chain of custody and that information just gets very lost and fragmented along the journey. So having this all taken in the cloud, having this all with the metadata surfaced and then having it in formats that you can reuse over and over again, makes the data really fair and makes it really available for everybody.

That's what we do. 

Kevin Folta: Well, I love the idea, but, but doesn't all this have to be curated in a way to make it organized where you can find what you're looking for? Because I think, you know, 90% of the data that we generate and maybe never even sees the light of day in publication because it, it leads us a different pathway.

It tells us something, you know, as an adjustment to the way we're doing things. Like, well, how [00:04:00] do you separate the wheat from the 

Mike Tarselli: chaff here? That's a fantastic question. So I, I'll say that first. We don't make a judgment call upfront. We don't say That's a really great error essay. That's a really bad result.

We actually take it all hoping that something downstream will help you to decipher that wheat from the chaff. And then what we do is when we surface that metadata and when we move it into this sort of vendor agnostic schema, a k a, like a data format, like an architecture, you know the way that, I'll give an example of a schema, a restaurant menu.

We all know that if you go to Egypt or South Africa or Japan or Argentina or the us you're still gonna have drinks and appetizers up front, mains in the middle, and desserts at the end, right? We've all agreed as a species that that's the schema for restaurant menus. So same thing, you, you can have a schema that's pretty darn consistent for a H P L C run or for a cloning vector or for you know, pcr R or something like that.

So what you do is you surface all those schema together. Then you get some real smart data scientists to look through them [00:05:00] all and find outliers, right? Either positive outliers, holy cow, this experiment did really well. Let's keep this data, write the paper, file the patent, whatever, or negative outliers, man, all these things crashed out.

Exclude those from the future analysis. But at least you have, you know, habeas corpus. You've got everything in your hands. 

Kevin Folta: Yeah, that's, that's good. I guess the other big question is where do you store it all? Because this is we have images that we take that are, you know, megabytes huge that. Thousands of them become gigabytes.

And you know, as we start to do more experiments, you, you see where I'm going, especially these time lapse experiments can just be massive. And so sometimes I think it's better, in my opinion, when we're doing the work, it's easier to redo the experiment than to save the, all the data from the previous experiments.

It, it, it just is so much data. Where does it go? You 

Mike Tarselli: know, it's fun cuz culture changes on the order of, you know, years to decades, not necessarily in days to weeks. I wish [00:06:00] it changed faster, but, you know come the 1950s and 1960s, everybody was still throwing their trash outside in rivers and, and, and burning things, right?

And, and they shouldn't. And now we've recycle that same kind of cultural shift is ongoing right now. The inflection points right now in biopharma, which says, Hey, Putting things on a hard desk or printing it out in paper or putting it on a USB or flash drive, whatever, and stashing it in a desk drawer is really not the way, the way is to put it onto a cloud of which there are several now publicly available.

We use Amazon Web Services or AWS behind the scenes, but there's a couple other alternatives. And the cloud has. I don't wanna say unlimited, but let's call it 99.9%, nearly limitless space because you're putting your data somewhere, you know, in a server farm in Nevada or Virginia or Dusseldorf, Germany or something, and you're putting it out there in a place that you literally could never fill up.

Now. Now your point about huge file size is still totally true. I believe that one of the [00:07:00] biotechs that came out big ICOs therapeutics in 2022 said they were generating five terabytes with a T of imaging data a day, and that was gonna be their cascade for a while. So, yeah. Where do you put all that?

Luckily the cloud is again, nearly limitless. It becomes more a question of how big a pipe you can afford and run, you know, efficiently, and whether or not you're capturing all that data on the way. Yeah, 

Kevin Folta: that, that's, that's a really good point. That always seemed to be the constraint for us was, was the pipe, you know, in terms of getting data up to that situation, especially when you're doing the analysis.

But let's talk about how. The evolution of the field. And we talked about this already, but what about you? I mean, it seems like you were the guy at the bench at one point. You have a background in medicine, you have worked in medicinal chemistry. And, and so what was your progression from that kind of work into modern lab 

Mike Tarselli: informatics?

Kevin, do you have time for a two second story? Sure. Okay. So, so when I was about five years [00:08:00] old, I was handed a sweatshirt that read Harvard on the front and my whole family said, Hey he knows how to read. He seems to like science. Maybe he'll be a medical doctor. Hooray, hurrah. And, and all my family are nurses, like everybody.

Grandmother, great-grandmother aunts sister-in-law, mother. Everybody's a nurse. Even even my daughter who's five years old, would love to be a nurse. So it seemed to fade a accompli that I was gonna go in the medical field. So I, I studied hard, you know did all my anatomy classes, did first aid, was a Boy Scout, all that stuff.

Worked a lot at, at, you know, trying to help the community, trying to get jobs early, volunteers, see what it was like to do that. And when I got to college, I entered pre-med. Side note, that's not a major. It's, it's a, it's a discipline, but it's not a major. You can't graduate with a degree in pre-med, sadly.

Which I didn't learn until later, but we'll get to that. And I took a class as an emt, so emergency medical technician. I said, I'll learn to be a doctor by going out there and helping people on the street or in the ambulance or, and figuring out if I like it. Well, it turns out I'm about [00:09:00] 6 1 2 50. So, you know, linebacker size, maybe not now, but linebacker size of the 1980s.

I, I faint when I see blood or sputum or oozing of any kind out of a body, didn't know that until I was standing in a hospital working on patients and then realizing I felt really bad in my stomach and then my head that is not a really good career progression for a medical doctor. Turns out can, can't do a lot when you're sitting there in surgery and go blah.

And takes a lot of orderlies to lift a person such as eye. So what I did instead is I said, you know, What do I love to do? What do I really like at the bottom of this? And I had a really great teacher in high school named Mr. Vito, and then another really great professor in college named Dave Adams.

These two gentlemen drew amazing biological structures on the board. They would draw up the, you know, surveys about heavy nitrogen that led people to figure out that, you know, where, how gene inheritance was done. They drew up the structure of DNA when, when you're like, you know, 12 and you don't really understand much, but [00:10:00] you know, you suddenly look at DNA N and you go, whoa, that's really neat.

How does that translate information? They drew up structures of, of minor proteins of, of molecules like caffeine, you know, well before I was probably supposed to learn them. And then, then organic chemistry itself, it's, it's an art of pictures. You learn biology through drawing pictures to represent molecules and then how they interact with the body, with receptors, et cetera.

So I just, I was hooked. I said, if I can't do the medical thing, I'm gonna at least help people some other way. So I got an internship in Boston worked at a company called Ariad Therapeutics for two summers. They basically plunked me in front of an H P L C and a large silica gel column and said, you're gonna be our separations guy for the summer.

I was like, yes, boss. Happy to do that. So did that. Eventually started submitting my compounds for assays, started making my own leads started, you know, running my own analytical runs. Started, you know, learning, designing, upgrading, and then eventually switched to Millennium Pharmaceuticals. Right.

This is right after the Human genome project. So they [00:11:00] were a wash in What if we could do personalized medicine all the way back in, oh man. 2000. 2001. Right. So they just had targets, dujour, and they said, okay, we're gonna hire you basically as an inflammation chemist. And I was like, okay, what's that do?

And they said, we don't know quite yet, but here's a series of drug leads. We need you to make 'em, you know, how does, I think the time my salary in, in 2003 was like $40,000? And I was like, I'm rich. 40,000 bucks. Amazing. So, you know, you're 23, 24, you don't know any better. You're, you're just gonna go out there and run the world on $40,000.

So, I did, I did, you know, missal chemistry for, for a long period of time, and then I hit a wall. I w I was looking through the biology as we, we had submitted one of my leads to early stage tox testing. It was being run in animal models. I was super excited and I wanted to help make decisions about where that molecule and that series went.

And I said, well, you can't. I said, what do you mean? And they said, well, you're, you're just a bachelor's holder. Like, you know, [00:12:00] PhDs are in these meetings, steering this stuff. Gotta get a PhD. I said, what? Like, how long does that take? Nobody in my, in my family does PhDs or MDs or stuff. They're all nurses, right?

I don't think my, my dad got went into school at all and my mom got a two year nursing degree, so I don't know what college is much less grad school. So I said, how long does that take? They said, I don't know, six, seven years. I said, I'll be out in four. Got out in four, called them back, said, I want my job.

They said, Ooh. We only hire people with postdocs.

You, you can see where this going. So, so I went out to Scripps and I did a postdoc for, for a year and a half and got a nature paper, ran a cool natural product no one had ever seen, put it in mouse models and realized it was nearly echo potent to morphine. Pretty cool right? Antinociceptive activity from something in the rainforest.

And I was like, wow, this is great. It's gonna really help my medicinal chemistry career. And unfortunately when I called back, it was 2009, it was the end of the recession. They couldn't hire anybody. Everybody was bricked up. I didn't get my job back [00:13:00] so long story short, I got into this field and, and kind of switched to informatics because at that time, the chemistry, biology, pharma roots all seemed closed to me.

So I said, what's still open? And the answer was data science and informatics. 

Kevin Folta: That's really funny because so many, you know, if I can give you a two second story too, please. It, it's, I've actually considered going into nursing now. Awesome. Because, well, because I've I have a parent who's not doing terribly well, and I've really appreciated the really effective nurses, and I've seen the ones who weren't so effective and I thought, you know, I would be more in the line of the really good ones and I could do it.

I could get a nursing degree in a year and a half of night classs go, and maybe my, maybe my next, you know, post. Whatever I'm doing now, career 

Mike Tarselli: university media mogul slash professor slash awesome personality. 

Kevin Folta: Well you know, I'm, I'm right. Nowadays, I'm not so much in research anymore. I've been kinda shuffled into teaching by the administration of a university, and it kind of has taken the, the passion out of what I do as a, as a researcher and[00:14:00] and so I think, you know, nursing would be kind of fun because on the kind of counter to how you feel about blood and guts is that when I see.

It hit the fan. I like lock down. I get this like, really intense focus and fix it and solve the problem. That's awesome. And I've been first person on the scene to three or four car accidents. Wow. Cause I live right by with a busy highway here and you hear it and I've pulled people outta cars. I've, I've had, you know, wow.

I've been front and center on first aid. You know, car batteries have traveled 300 feet down the road. And here I'm. Pulling somebody out who's stuff wedged under a dashboard, you know, it it th this has been it's something that just kind of works well for me, and I kind of think that'd be an interesting next, next, next chapter, you know, at, at this age.

But totally cool. And the other funny thing was, is that I got hired as an intern at a company where they, instead of giving me an h p LMC machine, they gave me a machine shop. Oh wow. And a, and a open checkbook at Ace Hardware, which was the [00:15:00] closest place at the time, in 1980, whatever, where you could get access to materials.

And so I would order lite and do all this stuff and build biotech equipment for them. And I would do that and then do the experiments. And we kept scaling up, scaling up, and then eventually could do these things that nobody else could do because we had massive throughput. We could analyze 4,000 seedlings at one time, you know, that kind of thing.

And this was all ISO assays and R F L pasting. Old school stuff, but they offered me a job out of a bachelor's degree for 35,000 a year. And I thought, sounds familiar. Whoa, this is more money than, what am I gonna do with 35,000 a year? And then they ended up but then I saw that glass ceiling. Yeah.

And then, and so I ended up going back for my PhD and took seven and a half years. No 

Mike Tarselli: comment. No comment. 

Kevin Folta: Yeah. But hey, the first, the first four years were pretty good, but the experiment couldn't be published because we couldn't figure out, we couldn't get by some technical impasse to make things interpretable in a way that were, was of impact.

But long story short you know, [00:16:00] very interesting. Twists and, and I was very much into informatics and computational biology until 2010 or until 2012 when I took an administrative position and then basically had my brain erased of how to do anything. So, and so, when I got spit out of the other end of that that monster and back to the lab I.

I kinda lost all my chops and, and the field had moved on so quickly in those seven years. Oh yeah. That, you know, what do you do? You know, where do you start again? So it, it, it, it's, it's funny how those two tracks are kind of parallel, 

Mike Tarselli: but, you know, I, I, I love that you said that. I'll, I'll share two things with you and then I promise we'll go back to the normal talking track is talking biotech.

However, I'll say that first, I, I learned from a professor long ago that there's three kinds of people in a university, really great administrators, Really great teachers, really great researchers, and it's really hard to mix and be two of any of those three at one time. Maybe you resonate with that, maybe you don't, I don't know.

And 

Kevin Folta: then no to, yeah, totally, totally, totally resonates. I, I was, I [00:17:00] was originally got into the administrative position because as an, as an interim, and they twisted my arm to do that. And I got into it and found out I had really good managerial skills. I really was good at dealing with people at, at understanding the psychology of a department as well as individuals within that department and found myself really helping everybody find their strengths and play to their strengths.

And it's awesome. It wasn't, it wasn't about me anymore, it was about all my faculty and the staff and the students and that was my main focus. And I still kept the research lab going. And and fully funded, you know, but it was killing me. And then still, you know, maybe speaking and teaching science, communication stuff, I, at least once every two weeks was on an airplane.

So I was just like, just dying. It was 18 hour days seven days, school week for about five, about five and a half years. And Told my boss, I gotta give up one of these three things. And turns out I ended up giving up all three. Oh man. Oh 

Mike Tarselli: man. Whew. And, and then the other [00:18:00] thing I would, I guess I would say is simply that though I love labwork.

Really do, I loved being in the bench. I loved being that person who washed my own glassware, prepared my own reagents, dried out my own catalysts, you know, did pre-preparation, et cetera. I, I'll be honest with you, I look back now. Basically last month there was published in the literature a technology that said we took all the common medicinal chemistry reactions and we put 'em on high throughput in 3 84 well plates, and you can run them in like serial dilutions and have like two or three of these plates run a week.

And I'm sitting here a GOG going, so that means me as a single operator. Could run over a thousand independent med chem reactions a week and get results. And purifications where when I was back at the bench lasted in, gosh, let's call it 2013, I probably maybe did 10 compounds a week. So that's two orders of magnitude since I left.

Kevin Folta: Yeah, so the big question with that in mind is how do we possibly train the next wave of scientists to be able to think in that kind of, that kind of level, at that [00:19:00] magnitude, I mean, you're talking about, you know, where we could work in a few test tubes in a, you know, in one of those racks where you could put 16 tubes across, you know?

Mm-hmm. 16 micro tubes across, now you're talking about 3 84 Well plates or even more Massive data sets that are computational. How do we get people on fire about dealing with such massive amounts of stuff? 

Mike Tarselli: Yeah, an excellent question. So, you know, I'm a big seven habits of highly productive people by Covey guy.

I think I used a Remember Borders hundreds of years ago. I used a Borders books gift card to buy that book back when I was like 18. And I opened it up and it really defined a lot of what I do with my life. So one of his habits is begin with the end in mind and. It's funny to say because you think that to a, like, let's call it an 18, 19 year old student who, who goes, well, I know what I'm gonna do.

I'm gonna get a four year degree and I'm gonna get an awesome job, and then I'm gonna get married, get a house, and then die. Right? That, that's the order, but that's not necessarily the career path you have to be on. You [00:20:00] and I just shared two very, I guess, antithetical or, you know different or hybrid career paths.

So, If your end in mind is you're going to have to deal with a thousand reactions a week, and then you have to assume the next five years, another order of magnitude, so let's call it 10,000. Then you buy rights, have to become data literate before you are science literate. That's, that's heresy, right? Like, I, I shouldn't be saying this aloud because all the professors of the world like, no, no, no.

You gotta learn your basics. Gotta learn your physics and your biology and your chemistry. Yes, comma. That's great it, but if you can't deal with the deluge of data you're gonna be presented with every single day from surfing through it and searching from defining what's real and what's not. From looking at validity, from looking at metadata, from looking at clustering and, and outliers and then like, we haven't even talked about AI or machine learning, right?

Getting to that, you need to understand very basic data literacy, so, Getting to the point where you can take a data set off of chem or zinc or PubMed, downloading it, opening it up, [00:21:00] realizing there's like 10,000 compounds there and a few thousand assays, and being able to surf through that and saying, okay, you know, what does a good result look like?

What does a negative result look like? You know, what do the header these columns look like? Can I organize and aggregate data? I do very basic statistics. Median mean mode, you know, confidence intervals, standard deviations. It sounds super dumb, right? Like you teach this to a, to a junior in high school now, but if you can't get a grasp on all that stuff, you, you have no hope of dealing with 10,000 reactions a week.

Right? No, you're exactly 

Kevin Folta: right. And but what's really interesting about this, and, and maybe you're saying this at just the right time, is that what you're saying about like, say scientific data applies to all information? Yep. And I'm teaching a class now that I, I, I woke up one day when I'm teaching some weird, I'm.

I actually was in my office getting ready for a lecture and I'm preparing for lecture, reviewing the pathways that I'm going to be teaching the students in the class. And I'm saying, if this stuff isn't second nature to me where I know this stuff like the back of my hand, why am I [00:22:00] possibly expecting them to know it?

Like it's the back of their hand And I realize that I'm giving folks. Esoteric details that are available everywhere, online or in textbooks. Yep. And, and so I switched, I mean, I had a hard switch and then now I'm teaching an undergraduate class on information literacy and, and so Cool. Yeah. What we're doing is we're seeing, how do we know.

What's real, what's not? What's good data? What's bad data? How do we vet it? How do we look at the metadata that are present in an article? How do we identify what's real and what's not? And a lot of work in lateral reading. A lot of cool concepts because all the data is out there, but. The problem is people don't know our filters are bad.

Yes. And we end up amplifying super bad information that's actually hurting people. You know, all the anti-VAX stuff, all the anti G M O stuff, it actually ends up with good people making the mistake of following a shiny pathway that has been polished to deceive them and. I need to help them sort that out.

And so it's funny that we talk about this in the [00:23:00] context, like you're saying in terms of scientific data, but that's just a microcosm of information in general, that there's a lot of shiny stuff that ain't so hot and other stuff that we have to be able to dig into and look at responsibly. 

Mike Tarselli: You got it.

Yeah. Yeah. And that's a wonderful thing you're doing teaching information literacy. And that's so needed. And I can't wait till that goes upstream down to like high school and middle school because, you know, guys, it's an information future. You know, it's the, the, the, the cat's outta the bag. The genius outta the bottle.

There is, there's no putting it back. The, the exabyte generation is upon us and they now say that the zettabyte generation is now coming. Right? Like we, we literally are starting to run outta prefixes. You know, I U pack and all the standards bodies literally invented two new prefixes last year to be like, well, I don't know, what do you call 10 to the 24th?

We gotta come up with something. Cuz right, the, the trend is going nuts. At Bio IT world this year I. Anastasia Christensen, super, very smart AI machine learning researcher at Pfizer said that 90%, [00:24:00] 90 0% of the information that we depend upon as an industry, biopharma and its ancillary industries was generated in the last two years.

Our, our, our, our brains, our, our monkey brains, our lizard brains are very hard to deal with. Exponents, as you probably well know. How can you possibly contextualize something where you're like, oh my gosh. So 90% of what we know and have generated came in the last two years. That means next year we're gonna create almost a whole new wealth of knowledge that we didn't even have before.

Kevin Folta: I can't even find two socks that match in the morning. 

Mike Tarselli: Yes. Much less which target reporter reacts in this assay with this molecule and this species. You know, like it's a lot, man. Yeah, 

Kevin Folta: it's, it's getting pretty wild. We're talking with Doc Dr. Mike Elli. He's the CS O of TetraScience. This is talking Biotech podcast.

We'll be back in just a moment. And now we're back on collabs talking Biotech podcast. We're speaking with Dr. Mike Tarsi. He's the c s O of Tetra Biosciences, [00:25:00] but also has been around the field and watched it change through time and has a lot of thoughts about how we best prepare for the future, both in training the future scientists as well as how we may be able to adjust ourselves.

And one of the things you've talked about is the resistance to adopting new technology because it's Not as approachable or maybe doesn't seem helpful at the onset, but how do you think we can overcome these barriers by improving that human machine? How do you, how do you think we can overcome these barriers by improving that human and machine 

Mike Tarselli: interface?

I. You know, it's funny because you look at the history of technology over time and nothing we have today though, it all seems very, oh, of course. We've always had iPhones and of course we've always had, you know, heads up displays in cars and electric cars to boot, et cetera. A lot of it had many false starts to start.

Right. General Magic right, was, was a firm that even long before the iPhone came out, tried to do a handheld device, so did Palm, so did Blackberry, so did a lot of these companies, and yet, Mainline cell phone use didn't take [00:26:00] off until like the late two thousands, right? So this thing that we're all holding in our hands, and depending on for our day-to-day existence, one of the most widely adopted technologies in history, didn't take off until, what, 15 years ago?

So, This, you're, you're looking at something that has to be probed with the culture over and over and over again. In this case, the human machine interface, right? We, we talk about the fact that ai, artificial intelligence, it's going to look at all the data we have, it's gonna mine it, it's gonna tell us what best drugs to make against which targets, which genes to pick to get the best resistant versions of soybeans or blackberries, you know, which, which things to put into which microorganisms to spin up that next amazing paint coating.

Yes, comma, it needs to be trained and loved and nurtured. And this AI boom is only the second or third AI boom since, you know, the term was coined, what, 70, 80 years ago. So we're, we're not gonna be there yet, if that makes any sense. Hopely It does. [00:27:00] I, I, I think that there is going to be a. Long curve of adoption where we take ai, which some, some article writers not me have un charitably referred to as a three or four year old learning how to walk.

You know, more than it's, you know, actually just this amazing genius wizard thing. You can do AI that trains on a single simple task pretty darn well, and where it has broad access to lots of information. Say for mapping technology or for natural language, or for predicting prices. Well, stock information is broadly available, right?

Price information is broadly available. People traveling maps have been doing it for, you know, 10,000 years. There's a large corpus to go on. Biopharma data. Pretty, yo, go on. So please. Oh, no, 

Kevin Folta: my, my thought on this started to look at the balance of the people making the data versus the people analyzing the data.

And when we look at either academic departments at universities or companies that are hiring, how much of this is going to be physical [00:28:00] lab space that's actually, you know, smashing the, the, the cell gu and, and generating the data versus the. Folks analyzing it, does it seem to be much heavier on the analysis side than on the production side?

Now 

Mike Tarselli: I have obviously a bias to admit here, which is I work for a company that loves to do the analysis and storage parts. So I have to say that that part's gonna grow. However, I will say even in the broader field, what you're seeing I, I, I remember the very first time, three, four years ago when I heard the term dry lab.

And I went, what are you talking about? Like, there is no such thing, you know, you're either in the wet lab or you're not because I, I'm a chemist, right? I'm, I'm dealing with solvents. I'm dealing with catalytic runs and bioprocessing and stuff. I, I don't do dry lab, but there's this encroaching term that means I don't actually work with.

Solvents or reactions or cultures. I stand back from the action and I have at my disposal all of the data output of this project. Oh yeah. And every other parallel project that's ever been done by our company, you know, and I'm going to look [00:29:00] for trends. I'm gonna tease out and subtle effects and see if I can really, you know, if you nudge that receptor with a little bit more, this what happens.

So that dry lab contingent is growing and it's growing by leaps and bounds. It is very hard for classic biopharma companies to hire from the best places now cuz they're getting snapped up by everywhere. You know, e every large internet startup, cloud startup, security government tech, defense, et cetera, all needs data scientists.

So biopharma has really turned to growing them instead. And what that means is they, they partner with universities. Or they develop their own internal academies for saying, okay, bench scientist who's been here for five, 10 years. You wanna be in data now? Great. We're gonna train you on Python. We're gonna train you on R.

We're gonna teach you a couple tricks with the data. We're gonna show you some data sets. We're gonna, you know, introduce you to Jupyter Notebooks or ipi or, or some other interactive environment, collab among them. And we're going to let you loose. You know, go interrogate your data. Go try making your own [00:30:00] fizz.

Go try, you know, enabling alpha fold or something. And, and they grow them from within. And those folks, they, they take off Kevin, like, like they, they go with through like half a year of training and then boom, they're put on a project team or three. And they are really an amazing additive effect. 

Kevin Folta: You know, that may be where I'm going too, if I don't go into nursing.

The 

Mike Tarselli: data science. Dry lab. Dry lab. 

Kevin Folta: Well, well, here's the funny part about that is that you know, that you know that I have this very gigantic box in a building that's full of really cool equipment and it's really hard to get the money to do anything with it these days and trying to get. The funds to do the work that I really wanna do in the things I find interesting is, is kind of tricky and we have some really cool stuff.

But I think that I could use that basic infrastructure to generate mountains of data that I could analyze in lots of creative ways if I got my tools back to be able to do that. And I, and one of the things that I've noticed is I've worked with a lot of folks in computational biology space who don't.

Have the biology side nailed down. So they're, they're great [00:31:00] computational folks. And, you know, we had one big transcriptome data set where someone said, you know, we gave them barcoded reads from different tissues to analyze. And one of the

Runs failed. You know, at the Illumina run didn't work. And someone came back to us and says, well, obviously in leaves there's no gene expression, you know, cause the, the, the, the, they couldn't, they, that was their conclusion because there were no reads in that batch of, of transplant. Oh, oh my. Yeah. Well, maybe we need, so this, this, but this is how How we need to have more people who are intimate with the biology getting into the data side.

So is that really the trend that you see in terms of how we should be training students and postdocs to at least have some good coding component for R and Python? You 

Mike Tarselli: know, we, we have a term here at Tetra that we call Cyborgs, and it's not Cy Borgs, like, you know, Jean-Luc Picard is Lauti or Arnold Schwarzenegger is the T 100.

It's more the sci borgs get it scientific cyborgs. And what we take from that is, People who can [00:32:00] speak nearly fluently, both the tech side, the I'm gonna deal with mountains of data and, and actually derive it into something useful. And the science side, Hey, I've worked in a lab before and I understand that leaves do indeed express genes.

Those people are rare still, but we have an amazing idea and feeling that they're going to become more and more needed and prescient in every field of inquiry that's tied to the life sciences. So what does this take, like you said, Either you gotta take a, a techie Ady in the wool CS guy or gal and you gotta teach 'em some biology.

You know, first the biology bootcamp like they do over at M I T and R I T or, and then, you know, bring 'em through a couple Coursera courses or something and then bring 'em to the point where they realize, yes, there is gene expression in every cell and leads are made of cells. You know, just to make those conclusions would be good.

Maybe even pharmas would want to give CS people internships more often so they can bring in and cross train. And then on the flip side, what you're talking about is more the academic scientists who are trained and steeped in the fields of biology, chemistry. [00:33:00] Physics need to be exposed to the data side early and often.

So what does this mean? This means probably, in my mind anyway, my, my own personal 2 cents, taking my company hat off. You should be giving them access to data sets early. Inviting 'em to hackathons, telling 'em about the career paths that are available here and that they can make a killing probably more than your 35,000 for my 40,000 that we just started the career with way more.

And that it's really useful. It's a really great way to make an impact on, on the world without even ever touching a, a cell or a person or, you know, a, a cell line because you can be looking at every single asset. Of the biopharma process from discovery through development, through clinical, through market, and then making decisions based on where the data guides you.

That's a really powerful position to be in. 

Kevin Folta: Kevin. Oh, very good. You know, just kind of a funny aside with, when we talk about that 35,000, you know what the first thing I thought of after I got the job offer and I went to the grocery store, I thought, I can buy al I'll be able to buy olives [00:34:00] and cheese, 

Mike Tarselli: all the finer things that I need.

Kevin Folta: I mean, it was, it was like, it would've, it was like moving from my $12,000 a year grad student job or into like a whole nother. It's, 

Mike Tarselli: it's funny you say that cuz same, I, I was I was living with my, my then girlfriend, now my wife, and I was saying to myself, you know what's amazing, I can now go and buy like a prime cut of meat or a nicer cereal because I can afford it.

Darn it. 

Kevin Folta: Oh, it's so, so funny how things change. So we, we talked about this idea of how you retool potentially retool people to be able to take on these other aspects of their career and be able to actually identify or work with data sets, but. How do we deal with the problem of good data?

Because you can spend a lot of time looking at bad data to run down a blind alley, and so how do labs deal with the issue of maintaining clean data and, and, and really be able to curate this in a useful way? 

Mike Tarselli: Excellent point. So I, I'll say [00:35:00] first that putting everything in a consistent schema is really critical.

And for those of your listeners who are maybe not used to these words, like schema and, and transformation and data pipeline, I'll simply define like I did with the menu analogy earlier ahead of the break. I'll simply say that you, you wanna put everything in a format that's easy, like, like a chapter book, right?

Has a, a cover and a title and a. You know, an index, and then it has chapter labels and it has a, a postscript and an appendix, right? People have been accustomed to that schema for books for hundreds of years. So think about that for your data. If, if you have, for example, let's take a data set that I'm really, really familiar with in N M R, nuclear Magnetic Resonance.

What do you do? You, you drop a sample into a giant magnet. You bombard it with radio waves and there's an induced dipole moment into the protons, and they start to, you know, communicate information to you about where they're lying in the molecule so they can tell you some stuff. Okay, what comes outta that?

A big wave, big induced decay wave comes outta that a [00:36:00] four a induced decay. It looks like ones and zeros and a wave amp amplitude that comes out the other side. So first you gotta transform that, you know, you gotta, you gotta transform that with a four a transform, and it becomes literally to be peaks and you gotta shim that.

You gotta move it and you gotta tune it, change the shouldering on the peaks and things, and you get this beautiful, nice, if you've never seen it before, it looks a little bit like the, the scribbles that happen on a, on a, on a wave form after you get an earthquake, right? A little bit like that. A across, you know, a zero to 12 PPM parts per million wave form.

It's great. You can get to do so much information about the structure composition stereochemistry of that molecule from just this one simple two minute experiment Now. What schema should that data be in? How should you capture that? What should you think about when you think about that? Well, what kind of stuff do you need?

You need some bucket to capture the raw wave. Okay, great. Here's the, the scribble at the beginning. The little, the little induced decay wave looks like a tornado put on its side. Then you need some sort of bucket to say, How'd you do the transform? And just, just a [00:37:00] reminder, you know, like, like a little, a little audit trail that says, Hey Kevin transformed this waveform on June 29th, 2023, doing this, this, this, and this with these, these algorithms and functions.

Great. Keep your, give yourself a reminder. Then you probably want to say, here's the techniques I use to smooth it. Correct the shoulders, all that stuff, and, and you know, put them in another bucket. And then you want a couple of things from the final spectrum you're gonna get, you want the locations of the peaks, so a list of the peaks, boom, boom, what comes out, how they're split.

How they interrelate. That seems pretty straightforward, right? But that S Schematization can be different if you're on a Bruker NMR to a variant NMR to a Jas O nmr, et cetera, that they're not done the same way. So if you can make those consistent first, put them in a common schema, then you can start looking at what good and bad looks like, right?

Good means it's got all the right requisite parts. It's got the right headers, the waveform is the right. Size, shape, amplitude. The peaks don't look all funky, [00:38:00] distorted. Like if you've got like a giant scratch, diagonal line across your random r well, it's probably bad. Or something didn't resolve. If it looks like a big smear, ditto, right?

You can look at those very basic things. You, you want a good use of AI in machine learning? Go look through Spectra and say, when something looks like trash, throw it out. They do that with images all the time. So once you've done that, you know you can start looking at finer details like, hey, this splitting constant should be seven hertz and I see 14 hertz.

Either that's an amazing thing that upends the laws of physics, or it's bad data. And then you just flag it up for a human observer to come in with some scientific intelligence and say, Hey, you know, check a look at this cuz I've seen 300 NRS today and this one looks really bad. That that's the first step I'd 

Kevin Folta: say.

That's one half of the equation, right? So how do we curate good data? And, and, and I appreciate your thoughts there about, you know, keeping this audit trail totally makes sense because if you don't understand what you're dealing with, then you can't interpret its next applications. But the other side of the coin is how do [00:39:00] we train people?

And we talked about this a little bit, but what's maybe the most effective way to be mentoring the next generation of scientists to be thinking about data and how does this really, you know, you mentioned in your experiences, I. That you had this, you know, volunteering run for a while. You mentioned earlier the volunteering that you did and, and how that helped craft your career and decisions.

How does having a service-based mindset really help us as professionals eventually in our fields? 

Mike Tarselli: I, I'll simply say that first, you don't know how to do everything nobody does. You know, the world's big and broad and, and, and you should go into it with a curious mindset every day of trying to learn and get better.

I know that I won't be paid money upfront for doing things until I can show some level of prescient. So what do I do? I go and do that work for free, and I know that there's a whole bunch of gurus and financial investment people who say, oh, no, never work for free. What are you doing? I. Say that that's not always true.

Sometimes the only way to learn something and get a new skill is to volunteer your [00:40:00] time, your effort, your curiosity, and have somebody teach you something for some hours. So in, in the field of informatics, I didn't learn by being paid to do it. I learned by saying, okay, I'm gonna sit down for a couple of hours and I'm gonna try to craft an HTML page.

Okay? I'm gonna try to learn to code in Python now and try to learn some very basic scripts. Okay? I'm gonna try to learn SQL and try to query a database and see if I can return certain values, even if it's something boring. Like the Timestamp Summit, an experiment was done at. You can get a very quick to your point about what's good data.

You can get a very quick idea from looking at data very fast about what's good and what's not. If you query for a string of numbers and it's time points and the numbers go 0 1, 2, 3, a, B, C, 4, 5, 6, well, you know what doesn't belong. That's pretty quick. And eventually you can refine those techniques through scripting, through recursive algorithms, et cetera, to figure out what really doesn't belong.

And you can have a machine do it for you. But anyway, getting back to your service mindset. You need to do these [00:41:00] volunteer projects to learn. You need to do, you know, if, if you wanna learn how to be a leader, you know, practice leading in a community group, in a religious group, in a boy scout troop, in a, you know, leading lab sessions for kids.

You, you wanna learn how to be a coder, great volunteer at a hackathon or, or go do one of those data sets like, you know, open malaria where you can, you know, learn on the data set. You can sort of make experiments and, and make mistakes in real time. But nobody worries because, you know. They're not paying you.

You, you wanna learn how to, you know, really be good and dedicated to a technique in a science. Well, you know, go try farm work or go try community gardening or go try, you know doing this work in a kitchen. All of these are gonna show you a hypothesis, a start, an experiment. Did it work or not? And then a reboot.

You know, like that's, that's what cooking is, right? That's what gardening is. That's what all these things are that you can learn to do in your own time. And then eventually your skills get good enough that people will pay you for 'em. That's, that's honestly how I [00:42:00] got into this world right now, and that's the ethos I use to mentor the next generation.

When people ask me for help I have a list off to the side. 250 some odd people that I've mentored in the past, sort of 10 years. I'm very, very, very proud and humbled by that. But, but most of them don't come and I don't like sit there and say, here's the secrets and keys to the universe. I ask them, how can you learn these things for free?

And who can you help? That will help you in turn by recommending you for that next position or giving you that insight? 

Kevin Folta: Yeah, that's outstanding advice. I, I agree with you a thousand percent. And it's amazing in parallels. I I, I've had so many students who show up cuz then they go back to the pre-med thing, you know, it's kind of like I need to have laboratory experience in order to get into medical school.

So you got those students who show up, they might come by, you know, check the boxes, you know, do a little bit of pipetting and call it a day. And then you get the other ones who show up and who sh will talk to you after class and say, I really want to get into this field. How do I do it? And I'll say, come on down, I'm passionate.

[00:43:00] Yeah, yeah. You can feel it. Yeah, you can, you can and say, come on down and, and well, can, can I meet you at seven o'clock rather than eight o'clock? Because I really want to get going, you know? 

Mike Tarselli: And, and those people are rare, right. You gotta breed that. You gotta cultivate 

Kevin Folta: that. Yes. Oh yeah. And I love that.

And, and, and, and I always ask the question, when I, when someone wants to get in the lab, I'll say, tell me about your high school job. And some of the ones who are, most of the ones who are duds will say, well, I really didn't do much. I was, you know, skateboarded a lot. And I, you know, maybe I was in a band and I, you know, I, I, I took the time for me because I really wanted to just have this before my pre-med career.

And then the other ones who come and say, oh, I didn't do anything special. I just was working in a restaurant. I, I was a cook. And sometimes I could pick up shifts as a busboy if, if they were available. And, and then I would work as a waiter sometimes, and then I, you know, if they needed somebody to do this, I was always willing to do it.

You know, that's the person you want. Amazing. Yep. And, and, and, yeah, so, so finding those folks who have that self-starting who aren't afraid of diverse challenges, who are the ones who are willing to pick it up and run with it. And I guess that's [00:44:00] how I was wired. So I kind of see that in others. And it turns out every single time those folks end up being the ones who stay in academia, who are leaders in industry.

And I have former undergrads who are CSOs too. And and it, it's and I agree with you. Thousand percent if you wanna play in this space, show up and say, how can I help you do what you do and learn something from it? And I don't need anything more than some mentoring here and there. And I, I, you know, and, and, and that's the way you really get a foot in the door.

Mike Tarselli: I, I could not agree with you more. I'm, I'm so glad we resonate on this, that we're copacetic. I'll, I'll tell you one more, ten second story, which is simply my very first job ever when I was 13, 14, was working for under the table wages with my then condo association. We were single parent household.

My, my mom, you know, never made too much money. But, you know, we, I, I helped clean up. I cleaned up dumpsters and I cleaned up on the grounds, you know, with a little picker and a bag and got paid like I think 10 bucks a week. But a lot of my stuff was volunteer. I didn't even report all my hours and in my second [00:45:00] job picked up more trash.

My third job, you know, picked up more trash. I was the facility kind of person and, and emptier of waste at my grad school lab. Same with my first startup, et cetera. So I've been the official janitor. So many times in my career, and you know what? I'm a C S O and, and you know what I do for fun when I'm not, you know, hanging out with my family or doing science stuff, I, I, I pick up trash for my local community.

And you know what, it's very rewarding. It does an immediate benefit. I'm not paid to do it, but it makes the place look a lot better. It's service leadership and, and people wanna hear what you have to say once you're past a thousand bags of trash. 

Kevin Folta: That's really funny. I, my, my first unofficial job, we just moved into a new neighborhood when I was a little kid.

And by 10 years old. And, and we were getting a lawn put in and they had three lawns in, in, in the three neighbors together that bought the grass from the same guy. And he was out there. He was an old man. He was out there putting down the pallets of sod by himself. And I said, what's going on? You know, shouldn't you have a whole team doing this?

And he said, well, everybody left. All the sods are gonna [00:46:00] die if I don't get it put down. And I said, well, maybe I could give you a hand. And he said, well you know, I can give you $2 an hour and you can, and you, and I'm thinking $2 an hour. This is like, you know, just hit a gold mine. So much money. Oh yeah.

10 years old. And I was thinking, I'm gonna go buy Kiss alive too when I'm done with this. Oh, so good. So, so, So I helped the guy lay sod and torture his summer sun. I looked like a lobster. I was so sunburned. I probably get skin cancer from it. Eventually, I was out there beating my brains out for this guy for I, I mean, three or four days all day, every day, lifting these giant rolls of sod and laying 'em out and getting 'em down on the thing.

And then at the end he's split and didn't pay me. Oh, 

Mike Tarselli: oh, Kevin, oh, I'm sorry. Can I mail you a check? 

Kevin Folta: You know what? It, it's come back in many ways, many times over. So I'm, you know, I, I think in the cosmic realm of things, that was kind of put me into into a very good receiving space because because I, you know, I don't know.

It, it just was, it was good. It was a good experience to get [00:47:00] shafted at 10 years old because you know, you realize that that. You know, the gratification was everybody got their grass down and I helped out an old man who probably would've died doing it on his own. And, and now maybe, maybe he 

Mike Tarselli: should've.

No, no comment. Hopefully we can edit that part out. No, I, I gotta ask you, cuz you brought up Kiss Alive too. I know you're a musician. I am too. Maybe I mean, I'm sure the people who are listening are like, come on more biotech. But, but tell me about your musical background a little bit and, and sort of, you know, how that's influenced your science, if at all.

Just curious. Oh, no, 

Kevin Folta: that's, that's, that's a funny story cuz I, I was started playing guitar with a friend of mine when I was seventh grade and really had a great time with it. And we did it mostly at recreation. We did, you know, played covers, did all that stuff and then but we always had an interesting.

A little edge to it. And everything that we did was a little bit different. And you know, the bands that we played in were all kind of funny and we had, you know, funny band names and, you know, we, we, it was in red Lobster Cult. We were in the blood worms. We, we had so many good ones. But the, the, the, the [00:48:00] pinnacle of that was Insane War Tomatoes where we had a very theatrical show.

We homemade pyrotechnics. That, that 

Mike Tarselli: we, I mean that's, that's why you're, that's why you're an agricultural researcher now, right? Because the insane war tomatoes. There you go. Well, and, and because 

Kevin Folta: we were using lots of fertilizer for NICs. We were, we were, we had stuff that we would, we, we set up at, at one university where we got hired to play on this gigantic stage.

And we, we thought, well, let's just try the pyro before we blew off all the ex. Explosions and stuff. And the guy who was running the date show was like, no, no, you will not do this. No, not this. If you do this, we're shutting it down. So we saved it all for the end and we, wow. We did an encore set and it blew up everything and it was really, really cool.

The thing that was, is that we were a little bit before our time. And a little bit too heavy metal for the punk rock guys and a little too punk rock for the metal guys. And so, and it was all you know, stage showing costumes and, you know, fire and slop and cannons that shot stuff out in costumes. It was really cool.[00:49:00] 

Umar would come along and do something similar five years later and and achieved a massive acclaim. So 

Mike Tarselli: yeah, so the AAR. Yeah, we used to run them in grad school, by the way. I I've heard them, I've heard Pixies, I've heard all these guys that famous monsters. By, by the misfits. I mean, we used to play this stuff in grad school all the time.

Oh yeah. 

Kevin Folta: That, that's, yeah. We were big, big, big influence by misfits and stuff and, you know, stuff like the sentence and all that. But but long story short is that after that I, I realized, and this was another thing you, I had to be, I couldn't continue muis, musician manship and be a full-time. Scientist, researcher, department, chair, podcast, or, yeah.

And so I, I stare at instruments and don't even touch them, but I I do stare at my seven week old daughter and realize that she has access to the most cool, old instruments that she'll have a good time learning, I hope. And so I, I have, I have big hopes that the next generation will. Well you know, maybe, maybe she doesn't need the homemade pyro, [00:50:00] gasoline fueled pyrotechnic show.

Mike Tarselli: Maybe not at seven weeks. 

Kevin Folta: Yeah, no. Well, the funny part was, is well back then was right after we got done doing all of this. Then the thing with Great White happened where they had the big fire and we looked at that and said, you know, that could have very easily been us. 

Mike Tarselli: So, so I, I grew up on the Massachusetts Rhode Island border, so that was huge news for us and we all, you know, took a page from that and said, you know, better OSHA rules safer shows.

You know, and you're right that that definitely changed the culture of sort of subculture shows. Yeah, that 

Kevin Folta: definitely did. And so, but that was what I did back then. And the fun part was I did do a lot of other stuff as kind of a, I guess a cheesy session musician where we had friends who would play on river boots and stuff, and they would call and say, we need somebody to, you know, fill in on bass tonight.

Can you make it? And they'd go, okay, yeah, sure. And then they'd gimme a cassette. And I would listen to the cassette on the way there about the songs they were going to play, and then could just kind of fake it all the way through. And then [00:51:00] other days they'd need me to play drums, or other days they'd need me to play guitar other days.

So it was always kind of a it was kind of fun and I'd get paid 50 bucks and got to hang out on a, on a riverboat for the night. And, So free drinks, good times. 

Mike Tarselli: I mean, Kevin, there, there you go. That's service, leadership and translational skills right there, because I'm sure that when you go get ready to go prepare for giving a talk, right?

You're doing the same thing. You're downloading the last three papers of that department or that, you know, host or whatever, you're skimming 'em on your plane, ride over to the talk. Then you're delivering your talk and trying to tailor it towards that audience. It's, it's very 

Kevin Folta: similar. Yeah, everything is last minute because for me it's especially even teaching a class, I can't sleep the night before.

I'm, no, I'm teaching because I'm trying to think of better ways to say what I'm. Planning to say, and I get more and more anxiety about that topic as it approaches. And it, so you're exactly right. The audience that I'm going to be talking to, who's in that audience? What do they need to leave with? What am I gonna, what am I leaving out that I really want them to take home?

[00:52:00] And, you know, those are the things that that. Drive me. How, how did I become the one being interviewed? Nice job. Flipping at that. I'm trying real hard, man. 

Mike Tarselli: Sorry. Old habits die hard. No, thank you. I appreciate so much, Kevin, that you're having not only me on here to have this discussion about, about data science, about informatics, about cloud, about, you know, next gen training, but that you allowed me to skew from the script a little bit.

Kevin Folta: Yeah. Well, let's let's do this again. Sometime I'll talk about my clown band maybe that I was in. I can't wait. Oh, we that's a great story because we, we had a band and we, our agreement was we're gonna dress like clowns and we're not going to tell anybody that it's us. And, and the best part was playing a show and having our friends tell us the next day about you should have seen what we saw last night at this clown band.

This was the most incredible thing we've ever seen. We're like, yeah, bullshit man. There's no such, so, so they didn't even realize it was us. That was so cool. Oh, that's amazing. But anyway. 

Mike Tarselli: Well now they know cause they listen to the podcast's. What? It's, yeah, 

Kevin Folta: but the cat's outta the bag. That's actually when, when you're in a clown band called [00:53:00] Amber Alert.

Oh 

Mike Tarselli: my gosh. Yeah, it was always, that's for the second 

Kevin Folta: podcast. Yeah, it was Evil Clowns. But anyway we'll go on from there next time. But thank you very much for joining me, Dr. Mike Elli, CS O of TetraScience. Thank you for joining me on Talking Biotech podcast and let's do it again sometime. It was a lot of fun and I bet if we do this in six months, the landscape of science will have changed.

Completely. So thank 

Mike Tarselli: you so much, Kevin. I appreciate it. 

Kevin Folta: Yeah, and for everybody else, thank you for listening to a rather interesting issue of talking biotech. Don't Google what I used to do. Thank you very much for, and you know, and by all means, do not report it to any universities or any, anything like that.

Thank you very much for joining me on The Talking Biotech podcast. We'll talk to you again next week.