The Computational Science of Drug Development

Jolie Hales:

The virus. Virus. The virus was here. Ernest, don’t get attacked by the virus. Hi, everyone. I’m Jolie Hales.

‍

Ernest de Leon:

And I’m Ernest de Leon.

‍

Jolie Hales:

And welcome to the Big Compute podcast. Here, we celebrate innovation in a world of virtually unlimited compute, and we do it one important story at a time. We talk about the stories behind scientists and engineers who are embracing the power of high performance computing to better the lives of all of us.

‍

Ernest de Leon:

From the products we use every day, to the technology of tomorrow, computational engineering plays a direct role in making it all happen, whether people know it or not.

‍

Jolie Hales:

Ernest.

‍

Ernest de Leon:

Jolie.

‍

Jolie Hales:

So much has happened to you since we last recorded. Oh my gosh. I mean, I don’t know. Do you care to fill our listeners in on the recent excitement in your life?

‍

Ernest de Leon:

Well, I think so. I mean, think they already know that my wife and I were expecting our second baby, our son. We have a daughter already. But what was unexpected was, when he came. He came six weeks early. And as a result, it was all kinds of chaos, because my wife made this list of all these things that needed to get done before he was born. And the doctor had told us, “This is his due date,” which was the end of May. So, we had planned all of this around, at earliest, two weeks before, because that’s exactly when our daughter was born. He was born six weeks early, so-

‍

Jolie Hales:

Oh man. What was the original due date?

‍

Ernest de Leon:

I want to say the 28th of May.

‍

Jolie Hales:

Oh my gosh. We’re so close together.

‍

Ernest de Leon:

And he was born on the 21st of April. Obviously, there were complications, so he had to be born early but the situation was, from my perspective, absolutely crazy. It was one of those where she had gone in to see her doctor for a normal appointment on that Thursday, just a checkup. They did some tests and they said, “You know what? Something looks a little off, and we’re going to send you over to the lab to get another test done.”

‍

Jolie Hales:

That’s never comforting.

‍

Ernest de Leon:

No. So she texts me and says, “Hey, they’re going to send me for another test.” And by the way, they did the same thing with our daughter so this was not out of the ordinary. So, she gets the test done, and she’d text me a couple of hours later saying, “They want to keep me overnight for observation.” This is the same thing that happened with our daughter, mind you.

‍

Jolie Hales:

Oh, really? I didn’t know that.

‍

Ernest de Leon:

Yeah, so we thought nothing of it. She texts me and asked me to bring her an overnight bag, basically. So I said, “Okay, no problem.” So I pack everything up. I head over to the hospital. During that period when I started packing that bag and arrived at the hospital, I walk into the room she’s in, and she tells me, “They’re going to take him today.”

‍

Jolie Hales:

Oh my gosh.

‍

Ernest de Leon:

And I’m sitting here like, “They’re going to take what?” And she’s like, “He’s going to be born today.” And so, the doctor’s in there. The nurses are in there, and the doctor’s like, “Yes, he needs to come out today. He’s a little early, but it’s not a big deal. He’s already almost full size,” all that kind of stuff. So, I’m sitting there thinking like, “Oh, Lord.” So we had already bought two sets of flights for my mother-in-law who was going to be the one to come help us.

‍

Jolie Hales:

Oh yeah, especially when you have the toddler at home.

‍

Ernest de Leon:

Right. The first thing I had to figure out was, what were we going to do about our daughter, because this was completely unplanned for?

‍

Jolie Hales:

Where was your daughter at this moment?

‍

Ernest de Leon:

At home with her nanny.

‍

Jolie Hales:

Oh, with the nanny, uh-huh.

‍

Ernest de Leon:

So I called the nanny and I told her what happened. I said, “It’s an emergency, blah, blah, blah.” She says, “Don’t worry about it. I’ll stay with her overnight. Do what you need to do.”

‍

Jolie Hales:

Go nanny.

‍

Ernest de Leon:

The best nanny. So then I immediately got on my phone and booked a flight for my mother-in-law to come the next morning. I was going back and forth and back and forth every day, like three, four times a day between the hospital and home, trying to keep both sides balanced. And of course, he was born early, so he had to go to the NICU and so it was insane and-

‍

Jolie Hales:

The chaos. And then you’re trying to figure out work, I’m sure, on the side.

‍

Ernest de Leon:

They said, “We’re going to put her in at 9:30.”

‍

Jolie Hales:

This is PM, right?

‍

Ernest de Leon:

PM. So I ran back home and grab the actual hospital bag that we were going to use for labor and delivery and grabbed the little plastic container I had and put some Texas soil in it.

‍

Jolie Hales:

Oh, that’s right. You did this with your daughter. Oh my gosh.

‍

Ernest de Leon:

Yes, and I ran back to the hospital. So I had everything now. I had the overnight bag, all the supplies we needed for all of that, put it into her room. And then I had that little thing in my pocket. Sure enough, 9:30 rolls around the doctor’s like, “Okay, it’s time,” and they wheel her in there and they’re getting her all prepped and they make me sit outside the door.

‍

Jolie Hales:

Is this for a C-section?

‍

Ernest de Leon:

Yeah, and so once everything was ready, then they let me in and then they do the procedure. So I come in there and I’m sitting there with my wife and I pull a little vial out of my pocket and I spread the Texas soil on the ground and just a little bit. It’s obviously an OR so you’re not supposed to-

‍

Jolie Hales:

I was going to say, what were the doctors thinking when you were doing this?

‍

Ernest de Leon:

They didn’t see it.

‍

Jolie Hales:

They’re like, “This is a sterile environment. Why are you pouring dirt on the ground?”

‍

Ernest de Leon:

It’s just a little bit of dust and it’s on the ground and it’s not going to harm anybody. But it wasn’t like I was bringing in-

‍

Jolie Hales:

Truckloads.

‍

Ernest de Leon:

… horrible bacteria and throwing it all over the place. This was just a little bit. So he’s born and it was probably collectively, if I look at that Thursday to let’s say the following, maybe Thursday, collectively the most chaotic week I’ve ever had in my entire life.

‍

Jolie Hales:

Oh man. Yeah. That would throw me for a loop as well. So wait, I have a couple questions here. I knew the story about you pouring the Texas soil underneath your baby, with your daughter, with your first child, so that she could be born over Texas soil but was she a C-section too?

‍

Ernest de Leon:

Yeah, she was.

‍

Jolie Hales:

Okay. That makes the story that much more hilarious because I mean, they must have seen you squatting on the ground for a minute-

‍

Ernest de Leon:

No, they didn’t.

‍

Jolie Hales:

… in the operating room, Ernest. For some reason, I always pictured this being in just typical labor and delivery where there’s all kinds of chaos going on. But in the actual operating room, while they’re cutting your wife open, you’re sprinkling dirt on the ground.

‍

Ernest de Leon:

Oh yeah, because I know that when something like that is going on, all of their attention is on the procedure. So they’re not paying attention to what I’m doing at all. So I can do essentially whatever I want.

‍

Jolie Hales:

What does this tube do? What does this go to?

‍

Ernest de Leon:

So I had it in my pocket and I just kind of reached in there and took a little pinch and did what I needed to do.

‍

Jolie Hales:

Oh my gosh.

‍

Ernest de Leon:

My wife saw me do it and she was laughing about it.

‍

Jolie Hales:

Oh, she did?

‍

Ernest de Leon:

Oh yeah, because she knew I was going to do it. There was no question about it.

‍

Jolie Hales:

So why over Texas soil? Why?

‍

Ernest de Leon:

It’s a concept of being a native Texan and you cannot be a native Texan if you weren’t born over Texas soil. That’s just how it works.

‍

Jolie Hales:

I think that’s so funny.

‍

Ernest de Leon:

Yeah. If push ever comes to shove, they were born over Texas soil. There’s no question about it.

‍

Jolie Hales:

That is the best.

‍

Sponge Bob:

Look Patrick. I’m Texas. Duh, howdy you all. I’m Texas too. Howdy you all. Get a dog, little longie.

‍

Jolie Hales:

So wait, so it’s been like three weeks since the baby was born. Is he home now?

‍

Ernest de Leon:

Oh yeah. He was in the NICU for a little under two weeks, then we brought him home. So he’s been home now for a while.

‍

Jolie Hales:

And how’s your daughter taking to the new-

‍

Ernest de Leon:

Oh, she loves him.

‍

Jolie Hales:

Really?

‍

Ernest de Leon:

Oh yeah.

‍

Jolie Hales:

Oh, that’s so sweet. That’s what we’re hoping for in our house too.

‍

Ernest de Leon:

Yeah. So speaking of all of this excitement and chaos and whatever, as I understand, obviously, because my wife and you were both on this pretty much the same schedule. You have your own exciting arrival on its way.

‍

Jolie Hales:

I do. And so far I’ve already made it past more weeks pregnant than your poor wife. I’m currently 37 weeks pregnant with a scheduled C-section at 39. So in less than two weeks. And just like you, this will be baby number two for us and for us, this is another boy. We have a boy who actually turns three on Sunday and then so it’ll be almost exactly three years apart.

‍

Ernest de Leon:

So have you all been taking a look at minivans and stuff like that?

‍

Jolie Hales:

It’s funny you should ask because right now we have this hatchback Hyundai Accent that I hate. It’s my husband’s. I’m like, “You’ve got to sell that.” But he loves that stupid car. And then I drive my Tacoma, but my Tacoma’s a 2004, so it’s pre the larger models. So instead of doing the practical thing and buying a van, we just bought a bigger Tacoma because I have always wanted a new Tacoma and I don’t know. It could hold the family right now so that’s how I justify it. My husband’s like, “Why don’t we get a van or a 4Runner even,” that’s like a Tacoma, but with more room for family. And I’m like, “Because I want to be selfish and I love Tacomas.” Did you guys get another car?

‍

Ernest de Leon:

No, that’s the problem. I have a Jeep. My wife has a car. It’s a Tesla Model 3. That’s actually what we’re using as the family car. And let me tell you, it’s a great car, but it’s super tight for us. And the cyber truck wasn’t out several years ago.

‍

Jolie Hales:

I know. I haven’t seen a single one on the road. I was asking my husband, I’m like, “Is that even available? Is that even an option?”

‍

Ernest de Leon:

No, it’s not available. It was meant to start going into production early in the pandemic. Obviously the pandemic hit, it threw everything back.

‍

Jolie Hales:

Kind of changed things a bit.

‍

Ernest de Leon:

The latest we’ve gotten from them is the first half of 2023 and so I pre-ordered that thing the day it got announced.

‍

Jolie Hales:

You did?

‍

Ernest de Leon:

Yeah. Yeah. I put down a deposit on it.

‍

Jolie Hales:

You put down a deposit on a cyber truck?

‍

Ernest de Leon:

Yeah.

‍

Jolie Hales:

That’s amazing. You should throw a rock at the window when you get it.

‍

Cybertruck Reveal:

Can you try to break this glass, please? Oh my! Well maybe, that was a little too hard.

‍

Ernest de Leon:

And I’ve been waiting for that thing. And that thing will easily fit the entire family because it’s a full size truck.

‍

Jolie Hales:

I mean, you’ll have to tell me when you actually get it and what you think of it.

‍

Jay Leno:

Pickup truck buyers tend to be pretty conservative in what they like. It’s needs to look like a pickup truck, that kind of thing. And this doesn’t look anything like a pickup truck, but immediately makes pickup trucks look old-fashioned.

‍

Jolie Hales:

Also, it’s been really interesting how having a second kid, it’s like with the first kid, we made an entire music video announcing the first kid. We had baby showers and lists of things and buying all of these things. And then with the second kid, I’m like, literally yesterday I said, “Oh, we should probably get some diapers and I bought some pacifiers and that’s it.” But I mean, it’s interesting how having a second kid really did change our vehicle situation. Just what you’re talking about, that was a thing that we had to pay attention to. And that was more expensive than all the stuff combined with the first one. But first kid gets a music video, second kid, basically a Facebook post-it note-

‍

Ernest de Leon:

Yeah. Same thing for us.

‍

Jolie Hales:

… announcing the second kid. And I probably should say too, there is a gripe I must express. So obviously I’ve always looked forward to meeting this cute little guy. But so a few weeks ago I was diagnosed with gestational diabetes and that basically means that I can’t eat any sugar and I can only seldom eat carbs. And I’ve got to tell you, Ernest, it is the worst.

‍

Ernest de Leon:

I can imagine.

‍

Jolie Hales:

Is the absolute worst. We’ve talked about my obsession over marshmallow peeps, and then I’m diagnosed literally a week before Easter and everybody’s eating them. But anyway, it’s been the worst thing for me in a first world problem since at least… I mean, comparatively in a global sense, not so much. But I really am such a whiner about this diet I have to be on because I’ve never been on a diet before. I’m a marathoner. I don’t need a diet.

‍

Ernest de Leon:

Yeah, no, I can imagine what it’s like having your sugar taken away, especially, it has to be one of those things where your body becomes chemically dependent on it.

‍

Jolie Hales:

And I definitely was. There’s no doubt.

‍

Ernest de Leon:

And then when you rip it away-

‍

Jolie Hales:

It’s like this withdrawal.

‍

Ernest de Leon:

Yeah, especially all of a sudden.

‍

Jolie Hales:

Right before Easter. What the heck?

‍

Ernest de Leon:

Right, instead of a gradual like, hey, we’re going to gradually ease you off this addictive substance. They just drop a hammer on you one day. That has to hurt.

‍

Jolie Hales:

Yeah, and of course, I’m being dramatic about it because I’m dramatic about things sometimes in my expression. But it’s just been really interesting because gestational diabetes, it’s different than a lot of other kinds of diabetes in some ways. So I’m still physically active in my third trimester, but because I have gestational diabetes, for whatever reason, pregnancy has decided to cause my body to develop this sort of insulin resistance. So normally if I eat sugar or carbs, then it goes into the cells is broken down, turned into energy. But because I now have this resistance, which comes from hormones produced by the placenta or something, I don’t really understand a hundred percent. So now if I eat any sugar or carbs, the sugar just stays in my blood and it’s not absorbed into the cells. It’s not converted into energy. And then when my blood is full of sugar, that sugar then passes on to the baby who doesn’t have insulin resistance.

‍

So the baby’s body converts it into tons of fat, and then it ends up making you birth a massive whale baby with an excessive sugar addiction, which I did the first time. That’s why I had to have an emergency C-section. It’s so funny. I was in labor for 36 hours and he wasn’t coming out. And they’re like, “Oh, well maybe he’s just a big baby like eight and a half pounds.” No, 10 and a half pounds, emergency C-section. He was a whale. Adorable but he totally looked like a sumo wrestler when he came out. And it was because I had developed gestational diabetes in that pregnancy, but it went undiagnosed.

‍

Ernest de Leon:

You didn’t have to go do that sugar drink thing, that test?

‍

Jolie Hales:

No, I did. They gave me that three hour sugar drink test.

‍

Ernest de Leon:

Yeah, yeah. That thing. Yeah.

‍

Jolie Hales:

And when I saw the results, I was googling it because I do that. And I could tell that even though I technically didn’t have it, I was extremely close.

‍

Ernest de Leon:

Very close, yeah.

‍

Jolie Hales:

A doctor could have called it either way, but they said, “Nope, you’re good.” And then they never tested me again, and that was probably 25 weeks or something. And so they should have tested me again a few weeks later, but they didn’t. This time, same exact thing happened but since I had had a whale for a baby the first time I was like, five weeks later, I actually asked for another test and the doctor said, “Oh, I guess we could do another one.” And it came back positive. What would’ve happened if I had not requested that?

‍

But just to give you an idea too, I’m 5’3″, 5’4″-ish, and I normally weigh 110 pounds. So, I mean, the first baby and all this stuff that came with it was basically more than 10% of me, or you could say at church, he was tithing for me and I looked like I was going to have twins, so I don’t know. It’s so funny that the doctors missed my gestational diabetes because I swear a photographer could have diagnosed me, but whatever, we’re all doing the best we can in life. And now he’s a healthy three year old, still addicted to sugar, but what are you going to do?

‍

Ernest de Leon:

Yeah, there’s nothing you can do. In my wife’s case, it was preeclampsia.

‍

Jolie Hales:

Which is also a common condition. That’s when your blood pressure goes up and then there’s no other way to cure it except to take the baby out.

‍

Ernest de Leon:

Right. Right. Also caused by something in the hormone or the placenta and whatever. So twice she’s had it. She had it with our daughter and that’s why she was born two weeks early and then obviously had it with our son six weeks early, her blood pressure was running high and they came back and said, “Yeah, you’ve got preeclampsia and it’s severe, so we need to take him out today.”

‍

Jolie Hales:

Oh man. It’s crazy,

‍

Ernest de Leon:

But it’s kind of weird how modern medicine has made the infant survival rate much better than what it used to be 50 or hundred years ago.

‍

Jolie Hales:

Oh, incredibly better.

‍

Ernest de Leon:

But the increase in the amount of conditions-

‍

Jolie Hales:

Has skyrocketed.

‍

Ernest de Leon:

… has skyrocketed, like you said. So it’s almost like the science innately knows that babies are going to need to be born earlier and earlier because of whatever condition is causing this. And so the science has to advance to the point where it can keep them alive. Like I said, our son was in the NICU and he was born six weeks early, but was of full weight and height for his time.

‍

Jolie Hales:

How much did he weigh? Can I know?

‍

Ernest de Leon:

He was almost six pounds.

‍

Jolie Hales:

Oh.

‍

Ernest de Leon:

But we saw some in there who were born at two pounds.

‍

Jolie Hales:

Yeah. It’s so crazy.

‍

Ernest de Leon:

One of them had been in there since December. So the fact that our son got to come home in less than two weeks was great.

‍

Jolie Hales:

And I’m so glad it worked out for you. And to me, it’s interesting how we have this fantastic technology, especially in the NICU, to take care of these babies who are two pounds, three pounds, but then we don’t have the medical technology to tell a pregnant woman she’s going to birth a 10 and a half pound baby. It’s like, did nobody just-

‍

Ernest de Leon:

Yeah, it’s coming. The technology is available now for a lot of this stuff, but it’s just not having to go through the FDA and all that, so-

‍

Jolie Hales:

Yeah, there’s a lot of red tape. I get it. The goal for this baby for me is, I don’t know, eight something pounds, at least not a full size whale. And there’s a point to all this TMI about my gestational diabetes, I swear. So in being diagnosed with gestational diabetes, I’m having to do the finger prick thing and then give myself these insulin shots and I have to take a look at all the medications that go along with diabetes and really consider what the diabetes is, how medications treat it. And I even sat down with my stepbrother who has type one diabetes, and I was so fascinated by how he manages it and also how before insulin injections existed in the early 1900s, people born with type one diabetes just didn’t survive very long, which is crazy.

‍

I mean, this condition has really opened up my eyes to this world that I hadn’t really known much about. And there are a number of medications used to treat diabetes. There’s insulin, which is considered the best treatment method in gestational diabetes because it doesn’t pass through the placenta to the baby, whereas other medications do. But there are other treatments that are popular in type 1 and type 2 diabetes, including drugs like, I don’t know if you’ve heard of any of these, metformin.

‍

Ernest de Leon:

My dad takes metformin.

‍

Jolie Hales:

Oh. And then I don’t know how to say this one, glimepiriride.

‍

Ernest de Leon:

Yeah, it sounds like it’s-

‍

Jolie Hales:

It looks like I just spelled that wrong in the script.

‍

Ernest de Leon:

Glimepiride or something. I don’t know.

‍

Computer:

Glimepiride.

‍

Jolie Hales:

And then another drug called canagliflozin, which is sold under the brand name in INVOKANA.

‍

Ernest de Leon:

INVOKANA, yeah.

‍

Jolie Hales:

I’m slaughtering these.

‍

Computer:

INVOKANA.

‍

Jolie Hales:

And then there’s a bunch of other ones. And in doing a bit of research on the subject, I actually got to talk to one of the people who developed that very drug canagliflozin.

‍

Ernest de Leon:

Oh nice.

‍

Steve Mehrman:

It’s an oral treatment. It’s a small molecule.

‍

Jolie Hales:

That’s Steve Mehrman, our undercover superhero for this episode.

‍

Steve Mehrman:

I was involved with the original piece when it first came in as a candidate development. And I helped build the process that scaled it up to metric tons. And then I changed positions that went in the, it was in solids dosage formulation development and developed the process and the formulation and scaled it up. So I was involved with that one almost end-to-end to make the tablets from the API, but we used a lot of fundamental modeling to design the studies for that as well as do the scale up. We did a lot of process analytical technology in line measures and making decisions about the process in real time. Also, a lot of batch modeling where we were looking at batch multivariate analysis projections to ensure that we had the right tech transfer in the product all the time.

‍

Ernest de Leon:

So it’s safe to say that Steve uses a lot of high performance computing to do this modeling.

‍

Jolie Hales:

Exactly. And this particular diabetes drug was created through Steve’s work with Janssen, which is the pharmaceutical part of Johnson & Johnson, who we’ve all heard of because not only do they develop a lot of beneficial medications and products, but they were also one of the three US companies to come up with a COVID-19 vaccine.

‍

News Clip:

This morning Johnson & Johnson out with new data that shows its single dose vaccine is working to provide long lasting protection.

‍

Ernest de Leon:

Yeah, the Johnson & Johnson’s a very big company. As a matter of fact, right down the street from me here, they leased a building and it says Johnson & Johnson Medical or something on their… Maybe it says surgical, I don’t remember. And I’m in Silicon Valley, so something to do with tech and medical in that facility. I’ve never been to it, so I have no idea. But I pass by it on the highway.

‍

Jolie Hales:

Uh, maybe it’s top secret.

‍

Ernest de Leon:

Well, I mean, I found it so it’s clearly not that secret.

‍

Jolie Hales:

Touche.

‍

Ernest de Leon:

And I found it by just driving by on the highway. It wasn’t like I was looking for it.

‍

Jolie Hales:

Yeah, there was a sign on the door, maybe they don’t have that many secrets going on.

‍

Ernest de Leon:

Yeah, big old sign on the top of the building, Johnson & Johnson Medical or whatever.

‍

Jolie Hales:

But it’s proof though. Johnson & Johnson is a very popular brand. It’s very well known in the United States and beyond. My sister actually and her family were recipients of that Johnson & Johnson COVID vaccine. So I mean, it’s hit my family personally in that way. And Steve actually helped develop that vaccine. So it’s interesting to see how medications like these can affect our lives so personally. And then for all I know, maybe my stepbrother is even taking canagliflozin. I don’t know.

‍

Ernest de Leon:

Maybe.

‍

Jolie Hales:

It’s possible. But before we dive into the kind of big compute work that Steve does for Johnson & Johnson, let’s learn a little bit more about Steve.

‍

Steve Mehrman:

I work in Janssen Pharmaceuticals in the biotherapeutics development area. I’m a director of data and analytics.

‍

Jolie Hales:

He’s been with Janssen or Johnson & Johnson for 23 years.

‍

Steve Mehrman:

I’m also an adjunct professor at Temple School of Pharmacy, and I’ve been teaching there for about seven years.

‍

Jolie Hales:

He also has a PhD in organic chemistry, and he’s always been passionate about applied data science to generate explicit knowledge. In fact, this is where in my conversation with Steve, we got into a bit of a deeper discussion about explicit knowledge versus tacit knowledge. And I don’t know, Ernest, are you savvy with the different knowledge definitions?

‍

Ernest de Leon:

No. I can make an inference from explicit and tacit, but I don’t know exactly what the difference is.

‍

Jolie Hales:

So in simplified terms, much more than what’s in Steve’s brainy head, tacit knowledge as I understand it, is knowledge gained from personal experience that is more difficult to express. For me, this would be knowledge of how to frame the best shot in a film or how to move the camera to compliment the tone of a scene. I feel like I just know how to do it based on my experience and the tone I’m trying to set. Whereas with explicit knowledge, that would be knowledge that’s more easy to articulate like something you can write down, you can share, maybe the answers to trivia questions or like a math equation or something. Or according to bloomfire.com, quote, “When data is processed, organized, structured, and interpreted, the result is explicit knowledge.” And then there’s implicit knowledge, which is the application of explicit knowledge.

‍

Ernest de Leon:

There’s a component of distillation and a component of extrapolation to all of this stuff.

‍

Jolie Hales:

If you say so. And as the director of data and analytics, I can see why Steve would gravitate toward explicit knowledge, which really comes into play throughout his career.

‍

Steve Mehrman:

Explicit knowledge in my mind, comes down to functional logic. It comes down to a model with an actionable output.

‍

Jolie Hales:

And Steve is hoping to push his field toward being as purely explicit as possible so the knowledge is completely operable so it’s obvious how you need to make decisions. And furthermore-

‍

Steve Mehrman:

My vision is to really enable the world’s best operator to run any process anywhere in the world 24/7, 365, through the use of data and modeling so we can generate that explicit knowledge to manufacture the best products for the patients.

‍

Ernest de Leon:

I mean, it makes sense. If you can get to the level of explicit knowledge, your ability to create something from that is much easier and much more refined or targeted than something where you don’t have that level of clarity.

‍

Jolie Hales:

Right. There’s more of a formulaic way of doing something. It’s written down. It just makes it more concrete. And when Steve isn’t pondering how to generate and record more and more explicit knowledge, he’s enjoying the great outdoors through walking, biking, hiking, and paddle boarding.

‍

Steve Mehrman:

I don’t watch a lot of TV. People always ask me, “Did you see this show? Do you binge this?” I’m like, “No.”

‍

Jolie Hales:

Which seems to be a common thread among science and engineering types. I mean, perhaps they can’t just sit and let their brains rest long enough to consume slow moving, fictional entertainment, I don’t know. As opposed to you Ernest and your fascination with… How do we put it?

‍

Ernest de Leon:

High end film.

‍

Jolie Hales:

Garbage movies.

‍

Lavalanchula Movie Trailer:

I saw spiders the size of men raining on hoods and spitting fire out of their mouths.

‍

Ernest de Leon:

I don’t watch regular TV either so-

‍

Jolie Hales:

Yeah, neither do we.

‍

Ernest de Leon:

And for me, it’s more of a when I see some of these shows on TV, if someone else is watching them, they actively make me angry because it is impossible for me to accept that there are people this stupid in our society. And it actually makes perfect sense why we have so many problems in the world today because we just have such a glut of morons in the world.

‍

Jolie Hales:

That’s such an optimistic viewpoint.

‍

Clip:

Your favorite dish?

‍

Clip:

My favorite dish? I like mugs because they’re very comfortable in your hand and they hold the hot things that you don’t have to touch.

‍

Ernest de Leon:

What really bugs me is every now and then Netflix comes out with something good and this happens all the time. Whenever a studio makes something completely brain dead, like a Michael Bay Transformers movie, it explodes.

‍

Jolie Hales:

It’s so true.

‍

Ernest de Leon:

Everyone is watching it. It makes billions of dollars.

‍

Jolie Hales:

But then the really good cerebral stuff is canned after two weeks because the general audience just doesn’t want to see that.

‍

Ernest de Leon:

Yeah. They can’t even process it. This is one of the reasons pretty much all I watch is YouTube.

‍

Jolie Hales:

I just watch documentaries.

‍

Ernest de Leon:

But high end film or-

‍

Jolie Hales:

Or garbage movies-

‍

Ernest de Leon:

… or pinnacle cinema as I call it, yeah, I love that. I still haven’t gotten to see Rubber. I’ve been trying to find some time.

‍

Rubber Movie Trailer:

Is what our killer looks like. A tire.

‍

Steve Mehrman:

In the back of my mind. I’m like, I think there’s something always more valuable that I can be doing.

‍

Jolie Hales:

Maybe he’d change his mind if he watched Lois & Clark. Oh, or Ernest, did you ever watch the series Sherlock with Benedict Cumberbatch?

‍

Ernest de Leon:

So I saw one or two episodes maybe, but I didn’t watch the whole series.

‍

Jolie Hales:

Oh.

‍

Sherlock TV Series:

Who are you? What do you do? What do you think? I’d say, private detective. When the police are out of their depth, they consult me.

‍

Jolie Hales:

So the series Sherlock is so good, except then it gets to season four and it plummets into a crap hole and dies a thousand deaths.

‍

Sherlock TV Series:

Come on. Be sensible. No, I don’t think so.

‍

Jolie Hales:

Maybe you’d enjoy season four-

‍

Ernest de Leon:

I might.

‍

Jolie Hales:

… come think of it. But anyway, going back to Steve and what he does participate in, Steve, actually, he dabbles in woodworking.

‍

Steve Mehrman:

I do quite a bit of cabinetry. I have probably four built-ins I built in my house that are just super fit for purpose. So I see the space and I’d be like, “What would be the most functional use of that space, almost, right, the coolest?” And then I draw it up and get approval from my wife, and then I go on a mission to build it.

‍

Jolie Hales:

And you can kind of see that logical mind and explicit knowledge at work in just how he even describes how he builds cabinets. I mean, he looks at a space and then he envisions the most functional use of that space. But anyway, so Steve works in the pharmaceutical industry, and I’ve always been curious about, for instance, COVID vaccine development. So we’ve talked to some COVID scientists in the past on this podcast who have used high performance computing to basically do a lot of amazing things and contribute to some critical breakthroughs. But I hadn’t yet spoken to somebody who had a direct hand in developing one of the COVID vaccines until I spoke to Steve. And Steve said that he first learned about COVID-19 toward the end of 2019, probably before most of us had ever really heard about it.

‍

News Clip:

China has identified the cause of the mysterious pneumonia outbreak in Wuhan City, and it’s from the same family that caused the deadly SARS epidemic 17 years ago. It’s a new type of coronavirus.

‍

Steve Mehrman:

I was reading a little bit of the global news and seeing some things, and you’re like, hmm, you got to pay attention. You don’t know what’s going to happen to it but I was thinking it might be something that would just come up and go away.

‍

Jolie Hales:

After all new viruses had come and gone a number of times over the years with some affecting the United States more than others.

‍

Steve Mehrman:

I kind of remember the moment of realizing, hey, something’s going on, but there’s been a lot of something going on for a while like mirror SARS, and you’re just like, it was regionally contained. So is this another piece of that, right? You didn’t know.

‍

Jolie Hales:

But then as 2020 started, weeks went by, then a few months, and suddenly the virus was here on our doorstep and everything was locking down.

‍

News Clip:

A SARS like virus, which has infected hundreds in China, has now reached the United States.

‍

President Trump:

Today, the World Health Organization officially announced that this is a global pandemic.

‍

News Clip:

The breaking news, stay at home. That is the order tonight from four state governors as the coronavirus pandemic spreads.

‍

Steve Mehrman:

It was surreal for me. I was actually on a golf trip with one of my buddies, and we were hanging out in a clubhouse after a round. The news was meh. Then about midweek it got crazy in March of that year. It was like a Hollywood script sitting there watching the TV, the news and the maps.

‍

Jolie Hales:

And soon life had veered onto a path that no one was familiar with.

‍

Steve Mehrman:

It was uneasy. It was one of those pieces you always like to be able to think that you’re going to know what you need to do, but I had no predictive model at that point in my head around what’s next?

‍

Jolie Hales:

And then there was his job. Remember, Steve worked on the research and development team at Janssen or Johnson & Johnson, a company and a team whose mission is to, quote, “Transform individual lives and fundamentally change the way diseases are managed, interpreted, and prevented,” close quote. Transform lives with things like vaccines.

‍

Steve Mehrman:

The team that I have just, we support the projects through that data systems and the analytics, the modeling pieces. So we support all the projects, all the different modalities. So I had a feeling we were going to get involved somehow because we have so many touchpoints.

‍

Jolie Hales:

And now that there was a new virus on the doorstep, threatening not just the lives of people across the globe, but those of their neighbors, their friends and their families, it was really personal.

‍

Steve Mehrman:

All the people that work in the pharmaceutical industry are people, and we have families and we all suffer from the same horrible diseases that everybody does. So we have the vested interest, but we also have the passion and the skillsets by design to help ourselves and everyone. So we’re in this together.

‍

Jolie Hales:

And so the company stepped into action. It was time to do anything they could to do what hadn’t been done before, create a safe lifesaving vaccine in record time.

‍

Steve Mehrman:

It really came out of some strong leadership that we had and passion for giving back, creating that product for unmet need. J&J’s always been philanthropic in this area so I think it was just the right time and right opportunity. So leadership was… We’re absolutely going to do this all hands on deck. And I think with combining that leadership drive with the fantastic domain expertise, passionate scientists, it’s a great combination to get stuff done.

‍

Jolie Hales:

A lot of the early work on the vaccine was done on a small team from Steve’s Pennsylvania location, while also collaborating with a talented group of people in the Netherlands after someone recommended Steve as this great resource to bring on board the project because he had this extensive knowledge and experience with computational modeling.

‍

Steve Mehrman:

I was in a meeting shortly after that and was basically recruiting a few people from my team. And we were embedded into several of the development workflows at that point to help really brainstorm even the potential for modeling. How can it help? How quickly can we get this done? And kind of triage the modeling opportunities to accelerate everything from designing the experiments to scaling up, to tech transfer, to how fast could we run this plant theoretically. So we were answering a lot of really cool questions.

‍

Jolie Hales:

Well, safety is always number one at Janssen, a critical factor in the development of this particular vaccine would be speed. I mean, the virus was already here and spreading faster than a grass fire and the world needed a vaccine right then. But in the past, vaccines took an average of, get this, 10 to 12 years to develop. I mean, before this point in time, the fastest developed vaccine of anything took around four and a half years and that was with a great amount of historical data already known and organized beforehand. But the truth of the matter was, I mean, with COVID, they didn’t have that kind of time.

‍

Steve Mehrman:

Safety’s number one, and what risks can you take? Where do we think that we can take the lowest risk, risk per se, and the fastest time that’s going to accelerate it? So, that’s really what it came down to so we leveraged a lot of the domain knowledge and really pushed the envelope on, if we had to accelerate this by six weeks, could we do it? Some things you can, some you can’t. You just had to design better studies, do them as quick as you could, learn fast and just run the next set of studies so you can de-risk as quickly as possible. So data played a huge role in that. Modeling and simulation played a huge role in that.

‍

Jolie Hales:

Modeling, simulation and high performance computing.

‍

Steve Mehrman:

We did a lot of computational fluid dynamic in the rescale environment to help accelerate that and design the studies.

‍

Jolie Hales:

And Steve’s approach is to simulate then run.

‍

Steve Mehrman:

I want to generate the best knowledge, design the best set of experiments before I physically go in and run them.

‍

Jolie Hales:

And before I go deeper into the COVID vaccine process, I do want to talk a little bit about how vaccines and other medications are made. Typically, the process starts with a discovery phase. At Janssen, there is a physical team dedicated to discovery, and this is the team that searches high and low for a potential core ingredient, otherwise known as an API or an active pharmaceutical ingredient, which is the ingredient in a drug that produces the intended effects.

‍

Steve Mehrman:

Discovery’s job is really identify that target and discover that modality to affect that target to create a positive clinical outcome, i.e. help people.

‍

Jolie Hales:

In fact, drugs are typically made up of two main components. So that API as one of them, the main ingredient that’s going to generate the immune response or maybe tackle some bug or whatever, and then the excipient, which is the non-drug substance that helps deliver the medication to where it needs to go in the human body. So excipients are typically not chemically active. They can be lactose or mineral oil or something, but it’s that active pharmaceutical ingredient, that API that really counts.

‍

Steve Mehrman:

We’ve got different ways of testing that receptor that they find or that target, and then figuring out if they can tune it in a way that’s going to help everyone.

‍

Jolie Hales:

But APIs can be difficult to discover. In fact, Ernest, do you remember at the end of 2020 when we spoke to Jerome Baudry about using high performance computing to do this thing?

‍

Ernest de Leon:

Right. His team was the one looking at natural substances that could potentially be used to help fight COVID.

‍

Jolie Hales:

Yeah, so in the episode, The Power of Plants to Pulverize Coronavirus, we spoke to a researcher who was heading up this very discovery process in hopes of finding, like you said, natural ingredients in substances like plants that could eventually become APIs in COVID therapeutics and he was using an HPE Cray Supercomputer to do it. He, through simulation, was able to run models to calculate how 200,000 different chemical compounds would potentially react with COVID-19. But he was quickly able to narrow down the 200,000 compounds down to just 125 that showed promise, and that was all done with computers.

‍

And from there, the 125 compounds were to be physically tested in a lab and then narrowed down to probably 5 or 10. In fact, I’m curious if any of those compounds actually panned out for Jerome. I really need to email him. I couldn’t find anything online. Anyway, so this is an example of the drug discovery phase. In many instances, drugs are discovered slowly without the aid of high performance computing and we talk about that in the episode too. But in Jerome’s case, the chemical compounds were narrowed down using high performance computing, shaving years and millions of dollars off of traditional discovery processes.

‍

Ernest de Leon:

Right and this is the exact same thing we see repeated throughout every industry. Things like Boom Supersonic running tons of simulations to get to a prototype that they think will function in the way they want rather than building-

‍

Jolie Hales:

All these prototypes.

‍

Ernest de Leon:

… thousands or millions of prototypes at millions of dollars.

‍

Jolie Hales:

Doing wind tunnels and hundreds of times over and yes.

‍

Ernest de Leon:

Yeah, this is the same thing that’s happening across all the industries. It’s not telling you exactly what the right thing is, it’s helping you eliminate tons and tons of what the wrong thing would be.

‍

Jolie Hales:

Exactly. Yeah, so this discovery process, so whether it’s done using high performance computing or physical experimentation, or maybe some combination of both, it’s part of creating every drug and vaccine. And at Janssen they use computational simulation and modeling during the discovery process as much as possible. And after a target API is discovered, it’s then passed to the development organization of which Steve is a part of at Janssen.

‍

Steve Mehrman:

They define what that, I’m going to call it an asset, is. Here it is. Here’s what you need to make. So they may have a very, very small, small amount of it, like micrograms, milligrams, but they’ve characterized it and they say, “Hey, you know what? Here’s kind of it. This is what it looks like.” And they’ve done some initial work on it to characterize it. So at least we have something we can be like, “Okay, we just need to make that.” Sure. So then it comes into development. And the job in development really is to build the best process at first launch.

‍

Jolie Hales:

So now that the active pharmaceutical ingredient has been identified, Steve’s development team is tasked with figuring out how to get that drug to interact with the human cells the way it’s supposed to. So how much of the API needs to be combined with what amount of excipient ingredient to be the most effective, whether that’s in a pill or a fluid or whatever?

‍

Steve Mehrman:

So really our job is to take that identified modality and make it for human ingestion and prepare it for commercial manufacturing.

‍

Jolie Hales:

And they use high performance computing to run models to figure out what path will be the most likely to succeed like you were talking about, eliminating the wrong paths. So how to make that drug the best it can be at the highest quality in the shortest amount of time possible. And they also need to figure out how to physically make a lot of the API. The discovery team may have identified that critical ingredient, but unless they have a way to duplicate it, there obviously just won’t be enough to go around.

‍

Steve Mehrman:

So how do we do that? Well, we leverage host cells for the most part.

‍

Jolie Hales:

Why try to make machines or something do the duplicating work when cells can do it for us?

‍

Steve Mehrman:

What we do is we codify these cells so that they can produce that protein for us. Because the cells, they’re great, they’re the best organic chemists. They can put together the fantastic construct of these proteins. So we let the machinery of that cell make the product for us. So what we need to do is understand how to make that cell make our product, and then we need to make lots of those cells making a lot of product.

‍

Jolie Hales:

In other words, the research and development team has to find the cells that will make the highest quality API in the best way possible. And then they have to physically duplicate those cells so that they have this sort of army of cells that can all make a ton of product or that important ingredient at the same time.

‍

Steve Mehrman:

We have a cell bay, so that stores those codified cells. So we go through this piece around cell line development, finding the best cell that we transfected with that codification to make our product. And then we screen through those and we find the clone that’s monoclonal that makes that product the best.

‍

Jolie Hales:

And once they find the ideal cell, they grow thousands and then millions of these cells.

‍

Steve Mehrman:

That’s our starting material. That’s the machinery. That’s the chef that’s going to make that product for us.

‍

Ernest de Leon:

Yeah, this is kind of the inevitable path of where I see humanity going in general, which is we are bioelectrical machines ourselves. We use electrical machines outside of us, computers to do all of this computation, high performance computing and everything but then we go back and use the biomechanical or bioelectrical machine to generate these type of things. So inevitably, I think what’s going to happen is there’s going to be a fusion at some point of what we would consider traditional computing electronics into humans, cyborgs essentially.

‍

Jolie Hales:

And biology, mind equals blown. Oh, and I should mention that while computational modeling comes into play a little bit during this part of the process, it is mostly done physically at this point, not quite in a mix of physical human tissue and robot put together like you’re talking about but that day’s coming, man. It’s coming.

‍

Ernest de Leon:

It’s close.

‍

Jolie Hales:

But once the project development team grows and expands these cells-

‍

Steve Mehrman:

Then we inoculate them in a bioreactor, so a vessel. Typically we start 250 mls or five liters in the development area.

‍

Jolie Hales:

You brush your teeth, you comb your hair and you inoculate Cellbags in a bio reactor.

‍

Ernest de Leon:

All in a day’s work.

‍

Jolie Hales:

Exactly. I mean, for those who work in non-chemistry based industries, a bioreactor is a closed container that is used to produce various chemicals and biological reactions, and they come in small sizes that can sit on a desk. In fact, I even looked on Amazon and you can buy a three liter bio reactor on Amazon. So there you go, everybody, jump on your app. They can also be huge vats that are lining the inside of a warehouse and they can be glass or metal and they usually have a bunch of tubes and stuff connected to the top that makes them look very scientifically laby. I mean, how’s that for a superb scientific description?

‍

Ernest de Leon:

Excellent. And nowadays you can pretty much get anything on Amazon. I would be surprised if you couldn’t buy, for that matter, a nuclear reactor on Amazon today.

‍

Jolie Hales:

Yikes. I didn’t check for nuclear.

‍

Ernest de Leon:

I didn’t either, but I’m just saying.

‍

Jolie Hales:

So once the development team has duplications of the best cells, they put those millions of cells into a bioreactor.

‍

Steve Mehrman:

What we want to do is create the best environments for those cells to produce our products. So then we go into a lot of engineering and modeling and science to understand how do we feed these? What do we feed these with, sensitivities that those cells might have? So that’s where computational fluid dynamics come in, that’s where it comes, all the omics, the intracellular omics to understand how that cell’s going to behave and how it’s the happiest. We need to make sure that there’s enough oxygen around it to keep it healthy and happy. There’s enough nutrients around it to keep it happy and healthy. So that’s where all that complex modeling really comes to play as part of the process development piece for the bioreactor.

‍

Jolie Hales:

In other words, Steve’s team uses computational modeling via fluid dynamics to determine how to treat the cells in the bioreactor so the cells succeed at creating that critical drug ingredient in the best form and timeline possible. And once they get the answers they’re most confident in, they then start running the physical tests based on their modeling.

‍

Ernest de Leon:

So it’s a cooperation of computational simulation and physical testing?

‍

Jolie Hales:

Exactly. First simulate then run the physical tests.

‍

Steve Mehrman:

We simulate up front and i.e. simulation in my mind up front is, I don’t know if you’ve heard of this old adage, it’s a day in the library is worth what? Two weeks in the laboratory. You can go to the library and help you design better experiments so that when you go into the lab and execute them, you don’t waste time trying to figure out stuff that people already knew.

‍

Jolie Hales:

Then after the cells have created that key ingredient in the bioreactor, the R&D team harvests the drug and then uses more modeling and engineering to figure out how to best clean, purify, and concentrate the product before passing it on to the next team in phase, which is drug development, where they figure out and develop the best physical way to deliver that medication to a patient so what kind of pill or solution and so forth.

‍

Steve Mehrman:

So this is the way that they find to stabilize, solubilize and deliver it to the patient.

‍

Jolie Hales:

And rather than just pass the baton to the product development team, Steve’s team continues to work with them using computational fluid dynamics via Rescale to best figure out how to proceed.

‍

Steve Mehrman:

There’s some really nice models that they have been using in a lot of the HPC to look at simulating how do we stabilize this? Or even solubility, pushing the envelope of solubility.

‍

Jolie Hales:

And throughout all of this, from discovery to product development, analytical development is helping along the way.

‍

Steve Mehrman:

They’re the ones that are developing the methods so that we can release the material in the future, understand the variance that we have in it, or what are some of the sensitivities that we have to stability, to process conditions that can affect the quality of the product. So there’s a lot of things that are very subtle and large molecules that require extensive analytical development.

‍

Jolie Hales:

Once the physical drug or vaccine is developed, it moves into the clinical testing phase where it’s tested in humans and when it makes it through that phase, it’s put on the market and hopefully starts really helping the population.

‍

Steve Mehrman:

It’s complicated because the human body’s complicated.

‍

Jolie Hales:

So complicated in fact that around 90% of drug candidates fail in US clinical trials, which is one of the reasons it typically takes 10 to 15 years and around a billion dollars to develop just one simple drug, or at least it has in the past. I mean, belief is that by continuing to utilize high performance computing wherever possible in this process, the timeline will and has in many cases shorten.

‍

Steve Mehrman:

So that’s a great area that computational work is going on right now that’s going to make a huge difference for the EIM and understanding, targeting and learning from the clinical.

‍

Jolie Hales:

So that’s the drug development process in a nutshell. And in the specific case of COVID-19 and the Janssen, Johnson & Johnson vaccine, this general drug development process was basically followed just under a tighter timeline with a few differences. For instance, because the virus was so new, new data were being discovered and released every day, but there wasn’t a lot of historical data available to allow them to crunch large data sets and know exactly what path to take right at first.

‍

Steve Mehrman:

We got generating on the fly.

‍

Jolie Hales:

And high performance computing really came into play when trying to figure out how the vaccine would interact with the human cells.

‍

Steve Mehrman:

We did use it a lot for that computational fluid dynamic piece to really understand sheer sensitivity, the optimal gassing strategies that we would have so that those cells that are making that product for it are healthy so they can make the most product the fastest.

‍

News Clip:

Unlike the Pfizer and Moderna vaccines, which use mRNA to carry the code to make the coronavirus spike protein which allows the coronavirus to invade human cells, the Johnson & Johnson vaccine uses DNA. The Johnson & Johnson vaccine also uses a modified weakened cold virus to gain entry into human cells. This then triggers an immune response teaching the body to fight off the real coronavirus.

‍

Jolie Hales:

And throughout this process with COVID-19 raging across the globe, Steve found that there was a unique spirit of cooperation and understanding on all levels, including with federal agencies, which is pretty unique. Everyone understood the urgency of the situation that a vaccine was needed and needed immediately and suddenly rigid old processes were reexamined in favor of new.

‍

Steve Mehrman:

The agencies were willing to work with us. For the first time they were willing to really move from what’s been written in stone for how we develop medicines. They really understood the need. So we really went down and focused on the core science and not the process, not the red tape. Why are we studying this when we know it has no risk, but we still have to file the paperwork type of thing. So I think there’s some pieces that we found some balance around where can we accelerate but still not have risk to the patient because safety’s always number one? But that was a huge evolution and I hope that, that’s going to really help us accelerate into the future as well.

‍

Ernest de Leon:

So the rigid old processes, they weren’t just ditched because somebody decided they were rigid and old, it was because they were now going to accept the risk of sidelining those because of the circumstances that this needed to happen. And I would not infer from this that this is going to be the norm going forward for normal drug research.

‍

Jolie Hales:

Maybe not, but it sounds like to me that a lot of the rigid old processes and then a lot of the red tape were based on old procedures and old way of doing things before technology had evolved. So there’s probably room to strike down some of that red tape and get rid of it. And this is my guess that maybe going through the COVID-19 vaccination approval process and cutting some of that tape, probably exposed some of those areas where it actually does make sense to make some cuts without actually risking the safety of others. And for eight months, Steve found himself pounding hard on the COVID vaccine project.

‍

Steve Mehrman:

We were having agile scrum approach that, okay, what’s the latest greatest that happened overnight? These experiments were run, here we’re at, here’s what we’re going to do, here’s the plan, here’s what reactor we’re going to be in, what can we do, answering those questions as quickly as we could.

‍

Jolie Hales:

And it helps since Janssen had developed vaccines before, they already had a pretty good idea of how to go about a vaccine development process for COVID just with a new element involved, that being the COVID virus itself.

‍

Steve Mehrman:

So it’s just a piece of codifying that part of that vaccine to recognize the COVID. So that’s a piece of it. So it’s a little bit of just almost engineering at that point. One of the things I was super excited about was just to see how fast we could turn around getting that sequence of that COVID variant and turn that into an applicable target. That was unreal, unprecedented from the discovery side and then pushing it into the pipeline.

‍

News Clip:

We start with good news this hour, minutes ago, Johnson & Johnson released results from its phase three trial of a COVID-19 vaccine, which could be approved for emergency use as soon as next week. It showed 85% protection against severe cases of the disease although the overall effective rate is lower. Some advantages, it is a single dose and it can be stored in a regular refrigerator.

‍

News Clip:

The vaccine was tested in more than 44,000 people in the US, Latin America, and South Africa. If authorized, it would be a much needed third vaccine available in the US to meet the demand that outpaces current supply.

‍

Jolie Hales:

By the end of 2021, more than 16 million Americans had gotten the Johnson & Johnson vaccine, the very vaccine Steve had been working on for months. And once the vaccine was out, Steve was happy to have fulfilled what he considered to be his duty before moving on to the next life helping project.

‍

Steve Mehrman:

It’s great that we could help the way we did jump in, but that’s why we’re here and I’m not one to toot my horn or whatever. So it’s your actions, what you deliver is the value back. So I thought it’s just as much helping myself and my family as it is to help everybody else

‍

Jolie Hales:

As far as HPC infrastructure is concerned for control systems that connect directly to the physical reactors and specific types of equipment, Janssen taps into an on-prem system and then for all of their modeling work, like all of their data sources, those are run on the cloud through Rescale.

‍

Steve Mehrman:

One of the things for Rescale for me was bringing MATLAB licenses out there. So that’s another thing because I think Rescale has been that platform that’s been almost host agnostic, which is great. And we have a lot of need for a diverse global team to have access to tools. And that’s one of the pieces of that concept of having this model factory where we have inception of model and then we build it into a product and then realize value. But one of the first things you have to do is reduce the barrier to get tools in the hands of the people that can do modeling. So that’s where Rescale has been great, let us host the licenses up there so that I could deploy those to anybody in the world, anywhere, anytime.

‍

Jolie Hales:

And Steve says they don’t really need an enormous amount of high performance computing to run their MATLAB workloads currently.

‍

Steve Mehrman:

But in the future, like I said, when we need that timeliness, we’re going to have to run it even faster. Five minutes might not be good enough. 30 minutes, we can’t let it sit there, go for four hours. If we could spit up a machine that can answer that question in two minutes, it’s worth finding that time to get that calculation done quickly so that we can react to that state, the process before it’s passed.

‍

Jolie Hales:

And when you have access to virtually unlimited compute on the cloud, you don’t have to rely on expanding and maintaining large and limiting data centers in order to scale up in that way. And I don’t think anyone will deny that computational simulation is speeding up the drug discovery process as it is with the process of developing state-of-the-art airplanes and fuel efficient cars and a long list of other products and discoveries.

‍

Ernest de Leon:

Yeah, absolutely. There’s no denying that computational simulation is rapidly evolving so many different areas of our day-to-day lives.

‍

Jolie Hales:

And the COVID-19 vaccine development process too was faster much, and thanks to computers. I mean, we already know by talking to other scientists before this episode, that high performance computing played a great role in modeling what the virus looked like and how it moved and acted and why it was so contagious. And then that information was shared with the global scientific community and then companies like Janssen set out on a mission to develop the vaccine.

‍

Ernest de Leon:

Yes, it’s very closely aligns with a previous episode we had talked about how when modeling tsunamis, they had to use a computational model that simulated an underwater landslide, and then that was fed as input into the model that showed the tsunami. So it was like one computational model feeding-

‍

Jolie Hales:

Into another.

‍

Ernest de Leon:

… in as parameters into the next one.

‍

Jolie Hales:

Totally. And Steve, who works extensively in computational modeling thinks there’s a lot further to go in the realm of digital maturity, which if you’re unfamiliar simply stated, digital maturity is apparently the ability of an organization to respond and take advantage of technological developments like computational simulation, cloud computing, just to name a couple.

‍

Steve Mehrman:

We’re still going through this digitalization maturity. It’s just something that’s not native to the culture. We’ve still got a little bit of, I think, paper based processes in place and acceptance of that’s just the way that it is we need to get away from and have better expectations. So we’re slowly going through that and it’s a culture piece too. But once we get through that and we have the data available, then it’s going to become the exciting piece. We can start making decisions faster, which means that it’s not three days later, maybe we can do it within an hour of running something.

‍

So we’re only an hour behind. For me, that’s atline. So then we’re pretty close, but not quite there yet. And then for me inline would be instantaneous. So if you take a look at the computational power that you’re going to need to go from offline to atline to inline, I’m going to be really interested to see what it’s going to take. Because if you’re generating high density of data, you’re going to need very high computational capability to return that result in 10 seconds versus 10 minutes or an hour. So it’s going to be an enormous game changer for us once this data starts flowing and we mature our ability to elicit explicit knowledge from that data.

‍

Jolie Hales:

Which just makes me think about how much more important access to enormous amounts of compute through the cloud will be when the pharmaceutical industry, in particular, has hit that level of digital maturity.

‍

Ernest de Leon:

And I think this is across the board, as every large company out there reaches that level of digital maturity, you’re going to see all kinds of things happening, improvements at every level from material science to just general operation of these random things. There’s a lot more to go in terms of how much more efficient and how much better we can make things with this type of compute power.

‍

Jolie Hales:

Exactly. And while technology, I mean, maybe advancing quickly, Steve believes that there’s a cultural element and a human component that needs to continue to develop as well in order to really take advantage of that advancing technology.

‍

Steve Mehrman:

So there’s an element to how do you make a decision from a model, which is a pretty big skill gap in my mind right now, and the ability to believe it and take action. So we have to propagate that not only through the core scientists and engineers at the bench, that we’d be changing kind of the way that they think about executing not just go in like, “Oh, I did this 10 years ago this way, this is the way I’m going to do it now.” Simulate then run, and then also propagating that through leadership that it’s worth trying to evolve to that state.

‍

Jolie Hales:

And going back to tacit versus explicit knowledge, Steve emphasizes the importance of computers granting us more explicit knowledge, more ability to learn and advance forward in drug discovery.

‍

Steve Mehrman:

When we were paper based, you never learned from anybody else’s projects. You’re not going to go read through their paper notebook. Even if you did, how are you going to remember and consume that? Without simulation, without modeling there’s really limited availability to learn from previous projects, current projects.

‍

Jolie Hales:

And Steve also emphasized that while having access to large amounts of compute power is important today, the need is going to skyrocket once personalized medicine becomes a reality.

‍

Steve Mehrman:

Personalized medicine is going to bring some significant challenges to the development process. You’re going to have a therapeutic area, then discovery is going to find that target, and then they’re going to define a modality to affect that target and that target and modality relationship in personalized medicine is going to go from one to many. So there’s going to be populations of patients where the target’s going to be the same, but how they react to that target is going to be tuned by that modality. So maybe there’s going to be five different subtypes of human population that we want to target so then that’s going to be five different therapeutic development targets that we need to develop.

Maybe they’re going after the same biological target, but then there’d be five modalities so we just increased our pipeline by five for the same clinical outcome, but we’re personalizing it into those buckets of the people so that we know that it’ll affect them better. So now we’ve got five types. That’s just one example. It may be more, it may be less, but we’re not going to be able to handle that with all the physical experiments we’re going to need to run. So we’re going to have to rely on more simulation, more learning across those different five development candidates.

‍

Ernest de Leon:

Yeah, and actually, let me tie this back to another previous episode we had where we talked a little bit about Mars exploration and a couple of other things. The same thing is actually happening right now with the next generation of rovers that are being designed. Historically, we did all of the heavy lifting and the computation on earth and we’re uploading things to these rovers. However, the next generations, plural, of those that are coming, NASA is working on being able to embed things like neural engines and HPC capability onto the rover itself, obviously not at the scale of what we’re talking about here with-

‍

Jolie Hales:

Like a data center strapped to the back of a rover.

‍

Ernest de Leon:

Like a data center, right, but enough to where the rover can, for example, take a 360 degree photo or video of its environment and have onboard computation to determine what is the optimal path for it to get from one place to another. Whereas before, it would take that information, that picture of that photo, send it to earth, we would crunch it on our side, determine what that path is-

‍

Jolie Hales:

Oh, the latency would be unideal.

‍

Ernest de Leon:

Send it back. It could be 24, 48 hours before the thing was even able to move. And then you ran into the situation of what happens if it moves, let’s say 15 feet in the direction you gave it. And there’s some kind of significant change in the terrain because of either wind or just natural movement of things, now you have to reanalyze again. So this is going to the personalized medicine thing. Being able to have HPC on a micro level as opposed to a macro level will open all kinds of doors in terms of science and technology for us in the future.

‍

Jolie Hales:

Oh, that’s so fascinating. And computational simulation and modeling in the case of drug development has come into play in much of the development that Steve and his team has done, not just with COVID but with drugs like the diabetes drug we mentioned earlier, canagliflozin, as well as the drug called REMICADE, I think is how you say it, which is a prescription for adults living with moderately to severely active Crohn’s disease. And as our listeners probably know, or many of them might, Crohn’s disease is a type of inflammatory bowel disease with no cure that is just not fun when left untreated so REMICADE can be helpful to patients in those situations.

‍

Steve Mehrman:

I’ve been involved with that one for a number of years.

‍

Jolie Hales:

He also worked on the drug called DARZALEX.

‍

Steve Mehrman:

There’s a lot of heroic efforts internally for that one.

‍

Jolie Hales:

DARZALEX is used to treat a type of blood cancer called multiple myeloma, where plasma cells basically grow out of control which can lead to low blood counts, bone problems, infections, kidney problems, and it comes with a five year survival rate of just 55%.

‍

Steve Mehrman:

That’s another one that was fast tracked and rightfully so. It’s fantastic treatment that went through our department and used a lot of the skills in our team to help with the data, help with analysis, scale up tech transfer. So we did a lot of work on that in short term.

‍

Jolie Hales:

Specifically, DARZALEX is a targeted monoclonal antibody that helps slow or even stop the progression of multiple myeloma by attaching itself to multiple myeloma cells in the body and then either directly kills those myeloma cells or allows the immune system to destroy them. And as someone who actually recently lost a loved one to cancer, I say anyone involved in cancer treatments is undoubtedly nothing less than a hero.

‍

Ernest de Leon:

Absolutely. And cancer’s another area where I think we are-

‍

Jolie Hales:

We’re on the cusp.

‍

Ernest de Leon:

We’re 50 years or less away from… not that there won’t be cancer, but that cancer will be able to be completely eradicated easily.

‍

Jolie Hales:

I think everyone listening to this is just praying for that day.

‍

Ernest de Leon:

Yeah, and so hopefully I see it before I die because likely I’m going to die of cancer, but-

‍

Jolie Hales:

I don’t like where we all are.

‍

Ernest de Leon:

Who knows at this point? Who knows?

‍

Jolie Hales:

Oh yeah. And as I was wrapping up my conversation actually with Steve, he did want to emphasize that science and engineering is a field really worth being in.

‍

Steve Mehrman:

I think it’s so fundamental and it’s a core piece of the innovation of this country and the world. We just need to have more people understand that the opportunities are endless and it’s at that edge of creativity. You can make yourself whatever you want to make in the future. So go out there, find out what you need to do and be the best at it you can be. And don’t be afraid of science and engineering, and it’s not putting you into a pigeonhole. So go out there, get a vision, and then find a way to make that vision happen regardless of what you do. But start with having a vision. Without a vision then you’re not driving towards something that you own.

‍

Jolie Hales:

Because that’s how these vaccines and life changing medications begin, with a vision. And I should say before we wrap this up, I know there are people out there who gripe about the pharmaceutical industry in one way or another, whether due to maybe costs of certain medications, or maybe some don’t work exactly as hoped on some portion of the population and whatnot, but I will say these scientists and these engineers working behind the scenes, they’re people just like you and me with families just like you and me. And I believe for the most part that they sincerely want to develop treatments to help all of us.

‍

Steve Mehrman:

We’re not only pharmaceutical employees, we’re patients.

‍

Jolie Hales:

And despite the imperfections of the industry, as with any industry, I’m grateful for those undercover superheroes who are responsible for the many, many breakthroughs that have not only helped make life better, but I mean have sustained it for many. And I think that Steve and his team falls into that category deserving of that gratitude.

‍

Ernest de Leon:

I would agree. There’s many areas where there’s always a balance. Yes, there’s always some kind of negative aspect to everything. And if you’re looking for it, you can find it but like you said, I think on the whole scientists and engineers are trying to better society and better humanity.

‍

Jolie Hales:

Right. And okay, let’s bring it back to my gestational diabetes, because I didn’t talk about it enough. So I didn’t have treatments for my first pregnancy as we discussed and the baby ended up being 10 and a half pounds, causing 36 hours of labor until finally forcing an emergency C-section, time in the NICU and then a longer than normal recovery time actually for me, among other issues. But this time around with treatments developed by blessed scientists and engineers, I give myself a few shots a day combined with monitoring my diet and hopefully it will result in a healthier baby and an overall better experience. So I guess I’ll keep you posted, but of course, if this baby is more than 10 pounds, again, I’m going to be really miffed that I stayed away from marshmallow peeps all this time for nothing.

‍

Ernest de Leon:

Yeah. I mean, that would be the greatest catastrophe, not having the peeps.

‍

Jolie Hales:

Hey, it is a sacrifice, Ernest.

‍

Steve Mehrman:

There’s almost nothing we can’t do. Just need to be creative and build it, build skills. I’d love to order like in Star Trek, my personalized medicine via voice command. That’s accelerated development in my mind. So shoot for the stars.

‍

Jolie Hales:

That final quote’s for you, Ernest, as a welcome back from paternity leave, a fellow Star Trek enthusiast. You’re welcome.

‍

Ernest de Leon:

Yeah, we’re kind of a, some would say, an annoying bunch.

‍

Jolie Hales:

Who would say that? Who?

‍

Star Trek Clip:

There’s nothing wrong with me. Maybe there’s something wrong with the universe.

‍

Jolie Hales:

To learn more about Steve Mehrman and his work, you can find him on LinkedIn. You can also learn more about Janssen by visiting janssen.com, spelled J-A-N-S-S-E-N.com. And you can learn more about parent company, Johnson & Johnson by visiting jnj.com, where you can read articles and watch videos about the company’s COVID-19 vaccine.

‍

Ernest de Leon:

Right, and I would like to point out that Steve is a merman as opposed to a mermaid.

‍

Zoolander Movie Clip:

I just thank the Lord she didn’t live to see her son is a mermaid.

Merman. Merman.

‍

Ernest de Leon:

You can also visit bigcompute.org to see pictures, videos, and a long list of articles that go into more depth on all the things we’ve discussed today. And you can support us by leaving a rating or review, or better yet, telling a friend about us and that would be on Apple Podcasts.

‍

Jolie Hales:

You can also leave a review and a rating on Spotify now.

‍

Ernest de Leon:

It’s right. You could do Spotify.

‍

Jolie Hales:

Well, actually, just the rating, I don’t think they allow reviews yet. And with that, the lawn mower has officially kicked on-

‍

Ernest de Leon:

Oh awesome.

‍

Jolie Hales:

… over here, which is totally good timing. So, I can handle a lawnmower at the very end.

‍

Ernest de Leon:

So here we go. Don’t forget, you can tweet both Jolie and I, we’re open to suggestions of all kinds. And don’t forget to use multifactor authentication and practice 3-2-1 backup.

‍

Jolie Hales:

Thanks for joining us today and stay safe out there.

‍

Ernest de Leon:

And don’t forget to mow your lawns.

‍

Jolie Hales:

I was thinking, I was like, man, God has answered my prayers. I’m in my closet, but whenever the landscapers come, it is just… I cannot record. But they came at the very end. Yay. I consider that a win.

‍

Ernest de Leon:

Yep.