Stream.Transcripts/twitch/1957952078 (2023-10-22) - Thad Hughes, AbCellera -- 22 Oct 2023/1957952078 (2023-10-22) - Thad Hughes, AbCellera -- 22 Oct 2023 -- Gigaohm Biological High Resistance Low Noise Information Brief.vtt

WEBVTT

00:00.000 --> 00:21.000
Test 1-2 testing 1-2 test test 1-2 testing test 1-2 test

01:30.920 --> 01:42.700
Gonna try and pack this all in in about a half an hour, because Mark is coming on at 8, and so I don't wanna get it his way, I'm just gonna kind of be a pre-show and give that the voice just a little bit of a rest

01:42.700 --> 01:59.700
The schedule for 16 minutes next is going on French, British, Italian, Japanese television.

01:59.700 --> 02:02.700
People everywhere are starting to listen to him.

02:02.700 --> 02:07.700
It's embarrassing.

02:07.700 --> 02:09.700
Kids deserve a lot of credit.

02:09.700 --> 02:12.700
This town's been flooded with phony twenties for weeks.

02:12.700 --> 02:14.700
Oh, it was nothing, really.

02:14.700 --> 02:20.700
But old Mr. Pietro posing as a doorman sure had us fooled for a while.

02:20.700 --> 02:24.700
He really gave himself away when he put on his little puppet show for us.

02:24.700 --> 02:26.700
The real hero of Scooby-Doo.

02:26.700 --> 02:29.700
By the way, where is he?

02:29.700 --> 02:32.700
Oh, no. Look at him.

02:32.700 --> 02:37.700
Like I said before, what a ham.

02:37.700 --> 02:42.700
Whoopee, whoopee, whoopee, whoop.

02:42.700 --> 02:47.700
God damn tea was too hot. I just burnt my throat. Dang it.

02:47.700 --> 03:08.700
Hello, everybody. Welcome to the show. Good to see you guys here. Thanks for joining me again tonight.

03:08.700 --> 03:14.700
Sound just a little scratchy, but I guess it's just something I'm going to get out of my throat here.

03:14.700 --> 03:19.700
No problems.

03:19.700 --> 03:25.700
I slept like a million dollars last night, which was just spectacular.

03:25.700 --> 03:29.700
If anything, my voice is a little bit lower today than it was yesterday.

03:29.700 --> 03:36.700
This is Gigo in biological and high-resistance local information free on how to stand my hours.

03:36.700 --> 03:41.700
2023, 22 of October.

03:41.700 --> 03:50.700
And we are still trying to fight through this limited spectrum of bait, which we have been trapped in by TV and social media.

03:50.700 --> 03:58.700
And we're still trying to make what is perceived to be true, actually true, Gigo's in general.

03:58.700 --> 04:03.700
Because as of right now, we're still misleadingly young, and that's going to lead to disaster.

04:03.700 --> 04:06.700
We don't fix that real soon.

04:06.700 --> 04:11.700
But they've been doing it to us for a while, right? That's why we all fell for it.

04:11.700 --> 04:20.700
And so anybody can come to life. It's never too late.

04:20.700 --> 04:26.700
And so we've got to make sure that our family and friends understand that it's not too late to admit that, you know,

04:26.700 --> 04:30.700
something screwy is going on and I should have known better.

04:30.700 --> 04:41.700
But it's getting more and more obvious that people have to actively ignore what's going on otherwise it's just obvious.

04:56.700 --> 05:06.700
So yeah, welcome to the show. Welcome to the show.

05:06.700 --> 05:16.700
We're going to do a short one tonight because, like I said, although I think I'm firing on all cylinders more or less,

05:16.700 --> 05:19.700
it still feels a little alien.

05:19.700 --> 05:24.700
And it still feels like I should be a little more careful.

05:25.700 --> 05:35.700
And so I'm not sure what to say about the obvious sort of settling in at an even slightly lower octave than yesterday.

05:35.700 --> 05:43.700
But it's been pretty consistent all day. And in fact, when I woke up this morning, if you can believe it, it was even lower than this.

05:44.700 --> 05:55.700
I don't know what to say. It's still, it's pretty close to the original instrument, but, you know, obviously things happened in my larnix or voice box or whatever.

05:55.700 --> 06:02.700
And so, yeah, we're just going to keep playing it by ear. I'm not, I don't feel like I'm stressing it at all.

06:02.700 --> 06:06.700
So it's, you know, just, I'm just going with it.

06:07.700 --> 06:12.700
Yeah, it was obviously inflamed. It was obvious. I mean, there's all kinds of things wrong with it.

06:12.700 --> 06:15.700
But now, I don't know, I'm just trying to keep my throat clean.

06:15.700 --> 06:20.700
I've been gargling some kind of colloidal silver mouthwash.

06:20.700 --> 06:27.700
I've been, you know, trying to brush my teeth a couple times a day, which I normally do, but now a little bit more vigilantly maybe.

06:27.700 --> 06:35.700
I don't know. I don't know what to do. I'm just trying to play it by ear and keep thinking because I'm definitely thankful.

06:36.700 --> 06:49.700
So don't take the bait on TV and social media. That's been our message for a long time now. And it is this Scooby-Doo mystery that they're still trying to get us to solve.

06:49.700 --> 07:03.700
And so, as we approach Halloween and then more importantly, Thanksgiving, when a lot of us are going to come together with our family and friends, it would be really good for us to have some, let's say, talking points ready to go.

07:03.700 --> 07:07.700
And so I'm kind of interested in input on that perspective.

07:07.700 --> 07:19.700
So if you, you know, you're that emailing type or you got an opinion that you always wanted to share in the next couple weeks, it might be a good idea because I think this is, this is one of those thanksgivings where a lot of progress might be made.

07:19.700 --> 07:22.700
So that's my hope.

07:22.700 --> 07:31.700
There are all these books that are out there that people might be talking about as they sit at your, you know, your, your dinner table.

07:31.700 --> 07:34.700
And so it might be good to already have something to talk about.

07:34.700 --> 07:38.700
And then this book apparently is going to come out on December 5th of course.

07:38.700 --> 07:42.700
I imagine it's going to be pushed back even farther because that's how this clown show goes.

07:42.700 --> 07:50.700
But hopefully it will be December 5th and so that we can, you know, make some progress from that, that sort of spiking of the football there.

07:51.700 --> 07:58.700
We're still trying to, I'm still trying to teach everybody that these people have been put into place.

07:58.700 --> 08:01.700
They were put in place a long time ago.

08:01.700 --> 08:03.700
I guess I get this music out of here.

08:03.700 --> 08:05.700
They were put in place a long time ago.

08:05.700 --> 08:09.700
And it's not something that a lot of us have been aware of for a long time.

08:09.700 --> 08:13.700
I mean, I, at one point during the pandemic.

08:14.700 --> 08:27.700
Thought that it was a borderline miracle that I finally got in touch with Brett Weinstein and that I was on a signal chat with him made it seem so like, you know, progress was really being made.

08:28.700 --> 08:30.700
And that's in the end.

08:30.700 --> 08:41.700
What we're, what we're dealing with here is that this is the control mechanism by which that Noam Chomsky quote in the beginning there is really executed this limited spectrum of debate.

08:41.700 --> 08:52.700
But then a lively discussion within that limited spectrum to make people think that there's a real genuine resistance, a genuine dissident, you know, opinion.

08:52.700 --> 08:57.700
And that's what this intellectual dark web does in a nutshell.

08:57.700 --> 09:10.700
And it's interesting to note that that that intellectual dark web was put in place around 2018 or 2019 largely by people like Elon Musk and Peter Teal.

09:11.700 --> 09:19.700
And then we now know that Peter Teal is invested in this antibody patent paradox to the hilt.

09:19.700 --> 09:23.700
It just becomes obvious that this is the story right here.

09:23.700 --> 09:34.700
The only question is, are we, am I fooling myself into thinking that what's on this screen is somehow revolutionary rather than exactly what they want.

09:34.700 --> 09:55.700
Because maybe the idea is to tear down the old system to expose it as garbage, but to play a shell game and get everybody to accept the idea that MRNA is a wonderful next generation replacement for this previously misguided intramuscular injection methodology.

09:56.700 --> 10:01.700
That's a distinct possibility that I don't dismiss lightly.

10:01.700 --> 10:03.700
And in fact, I don't dismiss at all.

10:03.700 --> 10:15.700
And I'm still trying to check myself to make sure that I'm not really just a puppet already here leading the charge to burn it all down or something like that, which is not really what I mean.

10:15.700 --> 10:19.700
But with regard to the vaccine schedule, I basically mean that.

10:20.700 --> 10:22.700
Let me see.

10:22.700 --> 10:23.700
Do I have this set up correctly?

10:23.700 --> 10:25.700
I think I do.

10:25.700 --> 10:29.700
So this illusion of consensus is what we're trying to break.

10:29.700 --> 10:31.700
It's these people that we're trying to pin down.

10:31.700 --> 10:35.700
And I'm going to continue to work on that on and off for the coming weeks.

10:35.700 --> 10:45.700
But right now, I think it's better for my voice and us that we do more of these study halls for the next couple weeks where, where maybe I'll take some notes.

10:45.700 --> 10:46.700
Maybe I won't depending on what it is.

10:46.700 --> 10:49.700
I'll just stop and talk about some of the things that are here.

10:49.700 --> 10:51.700
And tonight I have a very last night.

10:51.700 --> 10:52.700
Sorry.

10:52.700 --> 10:53.700
We had this.

10:53.700 --> 10:59.700
We had this show with with Annie DeGroote.

10:59.700 --> 11:03.700
And actually, that should be.

11:03.700 --> 11:04.700
Yeah, it should be root.

11:04.700 --> 11:05.700
Groot.

11:06.700 --> 11:11.700
And it's DeGroote, I think.

11:11.700 --> 11:16.700
And in Dutch, you would say, oh, long old for that.

11:16.700 --> 11:17.700
So that's DeGroote.

11:17.700 --> 11:18.700
Why?

11:18.700 --> 11:23.700
I haven't said a Dutch word since I broke my throat.

11:23.700 --> 11:25.700
So that's weird because that's actually back there.

11:25.700 --> 11:26.700
DeGroote.

11:26.700 --> 11:28.700
And now I can say.

11:28.700 --> 11:38.700
Anyway, Annie DeGroote is the leader of this epivax company.

11:38.700 --> 11:49.700
And so we were trying to, I was trying to take a little bit out of this talk about T-cell epitopes and a little bit about general immunology that I have been talking about before.

11:49.700 --> 11:53.700
That the T-cell epitopes were actually important, that they would actually look for them.

11:53.700 --> 12:07.700
And that there was this possibility that they would use T-cell epitopes planted in a biologic to sort of up or down, you know, up regulate the immunogenicity of an otherwise non-immunogenic protein.

12:07.700 --> 12:22.700
And so they could have done that with a coronavirus spike protein, which may not be very immunogenic because it has overlap with human sequences, which she showed in that discussion to be actually tolerogenic.

12:22.700 --> 12:28.700
So coronavirus proteins, especially the spike proteins, are designed to be tolerogenic.

12:28.700 --> 12:51.700
So if you wanted to make a coronavirus spike protein, one of the ways to do it, an epivax technology would have definitely been a way to do it, would be to take one of these or more of these small epitopes and insert them into a spike protein and then claim that this was the spike protein that you found on this novel coronavirus so that when you used it as an immunogen, it would already work.

12:52.700 --> 13:04.700
One of the other possibilities, of course, is to insert some novel sequence into the spike protein of the vaccine that you took, that you're basing on the viral protein.

13:04.700 --> 13:09.700
And either way, you would be doing the same thing, but it's ultra clever, right?

13:09.700 --> 13:15.700
If you put it in the wild sequence, and then when it's in the wild sequence, you just kind of shrug your shoulders.

13:15.700 --> 13:23.700
Well, guess it's there, even though all along, you knew that you had already designed it to have many of those epitopes and to be highly immunogenic.

13:23.700 --> 13:35.700
So that's one of those things where, again, this is where I tend to agree with people like Kevin McCarron because this is a very, was a fine line.

13:35.700 --> 13:44.700
And she was showing you this hot scale of almost like a skull ball rating of peppers, but she was looking at immunogenicity of proteins.

13:44.700 --> 13:54.700
And then talking about these small T cell epitopes and the more of them that there were, the higher on the immunogenicity scale, these proteins showed up.

13:54.700 --> 14:02.700
So that's extraordinary because then the next step there is, of course, can you make proteins that are more immunogenic?

14:02.700 --> 14:09.700
And her answer, of course, would be, who's asking? And if it was the DOD, she would say, absolutely, I can.

14:09.700 --> 14:23.700
And so here again, you just got to understand that although they have this technology deploying, it still would require recombinant DNA to make the RNA, and then that RNA would essentially be a clone, right?

14:23.700 --> 14:26.700
Because that's how you would do it. You've got to make it perfect.

14:26.700 --> 14:41.700
So there's, or the protein from it. But anyway, I wanted to watch this video because this video is another dude that works for Abselara, but he does the machine learning.

14:41.700 --> 14:52.700
And it's a short video. I can get it done by eight. And I'm going to stop it before it's all the way done because I really want to do is just get to the technology that they use to sort through these cells.

14:52.700 --> 14:56.700
So you can see the basic methodology isn't that magical.

14:56.700 --> 15:07.700
And then you can see really how actually it's possible that these guys just came together because of this microfluidics technology.

15:07.700 --> 15:15.700
And, I mean, glow on the dark beads and stuff like that is so basic technology that they use. It's actually quite extraordinary.

15:16.700 --> 15:28.700
And in a way, it's foolproof, but also, you know, you see again why all this talk that we've watched a couple of lectures of these people now.

15:28.700 --> 15:35.700
And they make these grave, great excuse me, grand statements about what they can accomplish and what they can do.

15:35.700 --> 15:42.700
And they have all this jargon and stuff. But in the end, it's just little tiny single cell wells that they're able to fill with single cells.

15:42.700 --> 15:46.700
There's little beads in there and there's some fluorescence and it's not really that clever.

15:46.700 --> 15:52.700
And that's, you know, maybe sounds arrogant from my perspective or from your perspective for me to say that.

15:52.700 --> 16:11.700
But I just, I don't want, I want, what I want you to see is that actually the technology to make that little container with microfluidic chambers is more important to this than the beads because the beads, I mean, it's, we're going to explain it when we listen to it.

16:11.700 --> 16:18.700
Let's just do that now.

16:18.700 --> 16:26.700
And I will shrink my head.

16:26.700 --> 16:32.700
I'm Fabius and I joined a seller six months ago and I'm now leading the machine learning team.

16:32.700 --> 16:35.700
My background is machine learning and software engineering.

16:35.700 --> 16:44.700
Before, I spent 10 years on Google speech recognition team, followed by another two years at Google health, working on computer vision for radiology.

16:44.700 --> 16:54.700
So first, here's a disclaimer about this talk. I'll give you a few seconds to read it over. This is from our lawyers.

16:54.700 --> 16:57.700
Next, my plan for today is to tell you about a seller is technology platform.

16:57.700 --> 17:06.700
Oh, so that, that part, just so you know that I didn't miss it, that part was just something where he, he has this statement up here that said this presentation contains forward looking statements.

17:06.700 --> 17:15.700
So he's just, he's basically saying that we're making a bunch of predictions, but as a result, the forward looking statements may not prove accurate.

17:15.700 --> 17:19.700
And so you can't hold this accountable of what we're talking shit about right now. We can't actually accomplish.

17:19.700 --> 17:27.700
So just, it's not some special disclaimer. It's just kind of sad, actually, because then he's admitting that this is just a sales pitch with a lot of tough talk.

17:27.700 --> 17:31.700
From our lawyers.

17:31.700 --> 17:41.700
Next, my plan for today is to tell you about a seller is technology platform for discovering therapeutic antibodies and how machine learning makes a key part of it.

17:42.700 --> 17:55.700
At the highest level, we at a seller are a tech company. We make long term investments in teams and technology so that we can fast track drug development and we partner with drug developers from large pharmaceutical companies to small biotech startups.

17:55.700 --> 18:00.700
We help them move quickly, reduce their costs and tackle their toughest problems.

18:00.700 --> 18:11.700
We built a modern technology platform for drug development and it ensures that the best science is quickly translated into new therapies so that we can benefit patients.

18:11.700 --> 18:20.700
Our arena for drug discovery is the adaptive immune system. And like all the biology, it's quickly becoming a goldmine for data science.

18:20.700 --> 18:25.700
Within each individual nature is operating at this massively diverse scale.

18:25.700 --> 18:35.700
Every one of us can make billions of different antibodies. It was almost like he kind of slipped there, right? He was going to say a goldmine and then he realized it all the way.

18:35.700 --> 18:42.700
You could see that he knew that was kind of like a double meaning. You know, it's a lot of money and also a lot of data.

18:42.700 --> 18:51.700
The ab in our name of Celera is a nod to these antibodies or abs as we affectionately call them. They're a crucial part of our immune system.

18:51.700 --> 19:04.700
This little green Y shaped symbol underneath represents the antibodies in this talk and it gives you a guarantee you during this entire talk he's not going to stop calling them antibodies is never going to call them abs.

19:04.700 --> 19:09.700
And that is just such a dumb statement. I can't. Anyway, I'm wasting time.

19:09.700 --> 19:15.700
Big idea of the shape of an antibody molecule.

19:15.700 --> 19:25.700
So at any given moment, we've got millions of immune cells making different antibodies and these antibodies form a molecular database based on the state of our health.

19:25.700 --> 19:31.700
They serve sort of as a memory of all the previous infection and disease challenges that we've ever faced.

19:31.700 --> 19:41.700
But one of the big problems in understanding adaptive immunity is the lack of technology to read this database to understand it and to change it.

19:41.700 --> 19:50.700
We can tackle this problem by analyzing the individual immune cells. Each of them produces one of the billions of antibody molecules a person can make.

19:50.700 --> 19:57.700
And every immune cell contains its own DNA code, which describes how to make a potentially therapeutic antibody.

19:57.700 --> 20:04.700
So the solution is to search through the immune system's vast database and find ways to interrogate one at a time.

20:04.700 --> 20:12.700
You have millions of cells that underlie our adaptive immune system.

20:12.700 --> 20:25.700
Appsellors drug discovery platform uses microfluidics, machine learning and other computation and customer robotics and automation to search, decode and analyze the natural immune system at record speed and depth.

20:25.700 --> 20:30.700
The speed and depth of our technology has redefined how antibody drugs are discovered.

20:30.700 --> 20:34.700
We've compressed a process that can take years into just weeks.

20:34.700 --> 20:39.700
And this is kind of giving you a bird's eye bird's eye view of what we do.

20:39.700 --> 20:52.700
It starts by sourcing a natural immune response, which can be in the form of a blood sample from a person or an animal exposed to a target pathogen, or from an animal immunized with some immunogen.

20:52.700 --> 21:01.700
The immune system can actually be directed to make antibodies against any target, such as those important in cancer or other targets that can cause other diseases.

21:02.700 --> 21:08.700
Next, we search the sample using our high throughput single cell screening pipeline.

21:08.700 --> 21:15.700
Here, we examine millions of immune cells to identify hundreds of high potential target specific antibodies.

21:15.700 --> 21:18.700
And this happens in just a matter of days.

21:18.700 --> 21:23.700
We use machine learning and computer vision here in ways I'll describe later.

21:24.700 --> 21:28.700
Computer vision. That's really funny. That's not.

21:28.700 --> 21:33.700
It's automated confocal microscopy and my guess is. But anyway, it's funny.

21:33.700 --> 21:38.700
Situate the genomic profile of the single cell antibodies we find within the broader content.

21:38.700 --> 21:41.700
Oh, my God. I just realized something, Mark. You're not going to believe this.

21:41.700 --> 21:50.700
But remember when Robert Malone was on the Tommy podcast with Hatfield? And I told you, he was on there.

21:51.700 --> 21:58.700
And he said, he said a completely different methodology for how he screened those drugs.

21:58.700 --> 22:05.700
So in one interview or more, one more than one interview, he said that he, it's okay.

22:05.700 --> 22:08.700
You can come through just in more than one interview.

22:08.700 --> 22:19.700
He said that he had made an x-ray crystallography model of the three CL protease and then interface that with all of the known drugs.

22:19.700 --> 22:23.700
And pharmaceuticals in the FDA catalog.

22:23.700 --> 22:33.700
And in another, in that specific Tommy interview, what he said was, is that they used high throughput confocal,

22:33.700 --> 22:42.700
my laser confocal microscopy, which is exactly, exactly what you would use in this scenario and what they're going to describe in five minutes,

22:42.700 --> 22:48.700
which is very, very curious because the two aren't the same.

22:48.700 --> 22:57.700
If you use a computer model of x-ray crystallography of an enzyme and interface that with computer models of different pharmaceuticals,

22:57.700 --> 23:06.700
that's entirely different than using live samples and high throughput laser scanning confocal microscopy, which is certainly how they're doing this.

23:06.700 --> 23:08.700
So that's curious to me.

23:08.700 --> 23:17.700
Next up, the overall immune response, which we can further sample using a process called repertoire profiling, and I'll describe that in more detail later.

23:17.700 --> 23:24.700
The most promising antibodies we find are expressed and further characterized based on approximately 500 different factors.

23:24.700 --> 23:35.700
And again, here we use machine learning, bioinformatics, and advanced data visualization techniques to determine their developability and suitability as drug candidates.

23:36.700 --> 23:44.700
So the result is a set of fully characterized and validated antibodies ready for development as potential therapies.

23:44.700 --> 23:53.700
As a side note, sometimes, when I talk to people outside of epsilon, one of the things they misunderstand is the difference between an antibody and a vaccine.

23:53.700 --> 24:04.700
So vaccines are designed to teach your body to make its own antibodies, but that process can take many weeks, and not everyone can survive an infection for that long,

24:04.700 --> 24:09.700
which is why you typically need to be vaccinated well in advance of your exposure to a disease or a pathogen.

24:09.700 --> 24:10.700
Yeah.

24:10.700 --> 24:13.700
But therapeutic antibodies can give you a shortcut.

24:13.700 --> 24:22.700
They give someone immediate access to a potent antibody long before their immune system could have figured it out, how to make something equally effective.

24:22.700 --> 24:26.700
So next, I'd like to walk you through more of the details of these steps.

24:26.700 --> 24:32.700
And along the way, I'll explain a few more ideas from immunology and also some of our technological innovations.

24:32.700 --> 24:38.700
And as you might imagine, there are many places on this pipeline where machine learning comes in handy.

24:38.700 --> 24:42.700
So I'll talk about those as well.

24:42.700 --> 24:46.700
I mentioned earlier that it all starts with a natural immune response.

24:46.700 --> 24:53.700
So something that's really fascinating to me is that all vertebrates have what's called an adaptive immune system.

24:53.700 --> 24:57.700
It starts to kick in within a week or two weeks of exposure to a pathogen.

24:57.700 --> 25:08.700
And unlike the rest of our body's cells, adaptive immune cells have special tricks to rearrange and mutate their own DNA as they try to create better antibodies.

25:08.700 --> 25:16.700
If you've ever implemented a genetic search algorithm on a computer, it's actually a pretty good analogy for what's going on with these B cells.

25:16.700 --> 25:22.700
Here, the fitness function is how well the resulting antibody binds to the target.

25:22.700 --> 25:29.700
So the B cells, what they're doing is searching for a protein structure that docks with a pathogen and binds to it.

25:29.700 --> 25:34.700
And the search is massively parallel. It's happening all over your body as the B cells divide.

25:34.700 --> 25:42.700
The result of the search, which progresses over the weeks and months following your exposure to a pathogen, is a huge diversity of B cells,

25:42.700 --> 25:45.700
each of which produces its own antibody protein.

25:46.700 --> 25:56.700
But since these B cells are scattered throughout the immune organs and the blood mixed with everything else, getting the right ones is like finding a needle in a haystack.

25:56.700 --> 26:06.700
So to look through the haystack, we developed an assay that contains hundreds of thousands of B cells in tiny isolated chambers on a microfluidic device, the size of a credit card.

26:07.700 --> 26:11.700
So one of the patents will be how they fill these cells with only one cell.

26:11.700 --> 26:18.700
There's probably a process or a robot or a series of solutions, whatever they do to get that.

26:18.700 --> 26:21.700
That would be one of the patented processes here.

26:21.700 --> 26:26.700
One way we can use it is to load individual B cells into the device's chambers.

26:26.700 --> 26:30.700
So here, the B cell is depicted as the green circle inside the chamber.

26:31.700 --> 26:40.700
And alongside the B cell, we can also load these tiny beads that are coated with different target antigen proteins, shown here as gray spheres.

26:40.700 --> 26:46.700
These isolated B cells in the chamber will then secrete their own antibody inside that chamber.

26:46.700 --> 26:56.700
So if the B cell is secreting a good antibody, an antibody that binds to the proteins on the beads, it'll stick to the beads that are coated with the antigen.

26:57.700 --> 27:00.700
Both the beads and the antibodies can be fluorescently tagged.

27:00.700 --> 27:04.700
And this results in a visual signature when the antibodies stick to the beads.

27:04.700 --> 27:08.700
We can pick up this signature using high-throughput fluorescence microscopy.

27:08.700 --> 27:10.700
And just as a side note...

27:10.700 --> 27:14.700
High-throughput fluorescence microscopy. You heard him say it.

27:14.700 --> 27:19.700
That's exactly what Robert Malone said on the Tommy podcast with Hatfield.

27:19.700 --> 27:20.700
I'm sure of it.

27:20.700 --> 27:25.700
Sometimes we also load other kinds of beads coated with proteins we don't want the antibodies to bind to.

27:25.700 --> 27:30.700
And again, whether or not the antibodies are sticking to those beads is also visible in the fluorescent...

27:30.700 --> 27:34.700
No, this is clever. And they would get to narrow things down a lot.

27:34.700 --> 27:38.700
But they're still left with hundreds of antibody candidates that bind their target proteins.

27:38.700 --> 27:45.700
So they could probably repeat this over and over with an ever more specific antigen set in here.

27:45.700 --> 27:50.700
But I don't think he's going to explain that totally here.

27:50.700 --> 27:55.700
So, as I mentioned, there are over 250,000 chambers in a single microfluidic device.

27:55.700 --> 28:03.700
And our high-throughput microscopy system images each chamber at several different wavelengths to create a multi-channel image,

28:03.700 --> 28:07.700
which is depicted on the bottom right in sort of a cartoon format.

28:07.700 --> 28:11.700
I have no doubt in my mind that this works for finding monoclonal antibodies.

28:11.700 --> 28:13.700
I have no doubt in my mind. It's a great idea.

28:13.700 --> 28:16.700
But now what they're doing is they're going to sequence these cells.

28:16.700 --> 28:18.700
Vision model models can help us.

28:18.700 --> 28:24.700
They're going to sequence these cells and then they're going to use those sequences in their informatics.

28:24.700 --> 28:26.700
That's really the trick here. You can see it.

28:26.700 --> 28:28.700
The particular binding profile that we want.

28:28.700 --> 28:32.700
For heavily multiplexed assays with many different types of beads,

28:32.700 --> 28:35.700
humans can't easily keep track of which bead types are binding and which aren't.

28:35.700 --> 28:38.700
So computer vision also enables richer multiplexing.

28:38.700 --> 28:41.700
See, machine vision, that's also just dumb.

28:42.700 --> 28:48.700
It is how you would do it in high-throughput fluorescence microscopy.

28:48.700 --> 28:54.700
You use different wavelengths and then you use a photon detector so your eye isn't seeing it.

28:54.700 --> 28:58.700
And it's using these words to try and bamboozle the audience.

28:58.700 --> 29:01.700
You're absolutely right. He's selling stock right now.

29:01.700 --> 29:07.700
But remember that a lot of how the momentum of the pandemic was created with this kind of stuff before the pandemic.

29:08.700 --> 29:14.700
This kind of momentum. We heard it in the 2019 talks that we watched two days ago where the guy was saying,

29:14.700 --> 29:18.700
he said anti-vaxxers. He said, take your flu shot.

29:18.700 --> 29:23.700
He said antibodies were the solution. He said vaccines were the only way.

29:23.700 --> 29:30.700
He said everything. So it shouldn't be underestimated how on narrative this stuff has always been

29:30.700 --> 29:36.700
and how it will remain on narrative that this technology is necessary to save us from our environment,

29:36.700 --> 29:42.700
from our mother nature. I've got about six minutes and then we're going to use a tonic live on YouTube.

29:42.700 --> 29:49.700
Our computer vision models use popular state-of-the-art architectures for classification, object detection and image segmentation.

29:49.700 --> 29:56.700
And one thing that's really important is they need to be robust and work across a wide variety of assay configurations

29:56.700 --> 30:01.700
and robust to the variance and noise that you typically see in these kinds of biological experiments.

30:02.700 --> 30:09.700
So once we've picked the hits using computer vision, we'll use an automated robotic process to recover them.

30:09.700 --> 30:16.700
And then we analyze the recovered B cells to identify the amino acid sequences of the antibodies they're producing.

30:16.700 --> 30:22.700
So what this lets us do is reproduce samples of the antibody for later characterization and validation.

30:23.700 --> 30:29.700
So one output from our bioinformatics pipeline is the genetic sequence from each B cell.

30:29.700 --> 30:35.700
And we can use this to figure out the amino acid sequences that make up each antibody.

30:35.700 --> 30:39.700
And this table is kind of a sketch of what that data looks like.

30:39.700 --> 30:42.700
So each row here is a single antibody.

30:42.700 --> 30:47.700
And each letter in the row represents an amino acid in the protein sequence.

30:48.700 --> 30:53.700
At the highest level, human antibodies are formed from two distinct amino acid chains.

30:53.700 --> 30:55.700
One is called the heavy chain.

30:55.700 --> 31:01.700
And it's shown here spanning most of the slide, the width of most of the slide, with a dark gray header.

31:01.700 --> 31:05.700
And on the right, there's a light chain with a lighter gray header.

31:05.700 --> 31:10.700
I've actually chopped out many of the heavy chains amino acids and all the amino acids from the light chain

31:10.700 --> 31:12.700
so that this table would fit on the screen.

31:12.700 --> 31:16.700
But what I want you to see is that each chain has a few different regions within it.

31:16.700 --> 31:22.700
The blue regions are called the framework regions and they provide a scaffolding structure for the antibody.

31:22.700 --> 31:25.700
And they vary less across different antibodies.

31:25.700 --> 31:28.700
Whereas the orange regions are known as the CDR loops.

31:28.700 --> 31:32.700
And they're the part of the antibody that actually determines what it's going to bind to.

31:32.700 --> 31:35.700
And so there's an incredibly rich diversity here.

31:35.700 --> 31:43.700
If you remember the idea of the genetic search, this is where most of the point mutations in the immune system search are focused

31:43.700 --> 31:46.700
is on the CDR loops.

31:46.700 --> 31:51.700
So at this point, what we've done is we've gone from an enormous haystack of images

31:51.700 --> 31:56.700
to a smaller but still daunting haystack of amino acid sequences.

31:56.700 --> 32:00.700
And as you might guess, it's hard to just read these sequences directly.

32:00.700 --> 32:04.700
But there's been an explosion of techniques in the field of natural language processing

32:04.700 --> 32:09.700
that fit perfectly onto these kinds of data sets.

32:09.700 --> 32:14.700
So you heard him in the beginning, right? He worked for Google and he did language processing AI models

32:14.700 --> 32:20.700
or machine learning models. And now he's going to apply those to understanding the antibodies.

32:20.700 --> 32:25.700
And what you're going to see is an interesting thing.

32:25.700 --> 32:28.700
And then we're going to let it go. It's like three minutes and then we're done here.

32:28.700 --> 32:34.700
One example, we can approach the complexity of this sequence space using dimensionality reduction.

32:34.700 --> 32:39.700
So what this means is we take the high dimensional amino acid sequences for each antibody

32:39.700 --> 32:42.700
and encode them into a lower dimensional embedding vector.

32:42.700 --> 32:48.700
So what you see here, each point in this plot is a single antibody heavy chain.

32:48.700 --> 32:55.700
And we've projected the sequence space down to just two dimensions so that we can plot the antibodies on the screen.

32:55.700 --> 33:00.700
And each of the three panels here shows the sequences that came from a different donor.

33:00.700 --> 33:05.700
And what's really nice about the way this turned out is that the 2D projection of the sequences

33:05.700 --> 33:08.700
isn't really clustering the donors away from one another.

33:08.700 --> 33:15.700
It could have turned out that the differences between different donors were so large that each donor would land in its own region

33:15.700 --> 33:18.700
of this two-dimensional projection space. But that didn't happen.

33:18.700 --> 33:27.700
Instead, what you see is that each donor's antibodies are well distributed across the entire two-dimensional projection space.

33:28.700 --> 33:32.700
What's cool is that if we color the antibodies using some metric of...

33:32.700 --> 33:42.700
It's not really clear to me what this two-dimensional space is and what the heavy chain U-map and whatever one and zero is.

33:42.700 --> 33:44.700
And so that makes it somewhat confusing.

33:44.700 --> 33:49.700
But let's just assume that they're using some uniform classification.

33:49.700 --> 33:56.700
And you can see that the cluster patterns across these different donors is pretty similar, which is interesting

33:56.700 --> 34:04.700
in that, again, what we saw with epivacs, there are likely certain residues in these proteins

34:04.700 --> 34:11.700
which attract the attention of the immune system for whatever reason being immunogenic, even across donors.

34:11.700 --> 34:22.700
And that should lead you to believe that this lack, almost lack of diversity here is because to a certain extent,

34:22.700 --> 34:31.700
there's a limited palette of foreign and immunogenic sequences that not surprisingly is pretty similar across humans.

34:31.700 --> 34:40.700
I mean, we use different HLA subtypes, but the diversity within that subtype from a given pathogen shouldn't be that different.

34:40.700 --> 34:46.700
What should be different is how what antibody sequence is chosen to represent that portion.

34:46.700 --> 34:55.700
And so it's curious that they cue in on the right on the same epitopes irrespective of whether they have the same HLA subtype.

34:55.700 --> 35:02.700
So what I believe this suggests from the perspective of the other technology that we saw the other day with epivacs

35:02.700 --> 35:09.700
is that the HLA specificity changes certain amino acids in the projected epitope and certain amino acids,

35:09.700 --> 35:14.700
which are known for in the selected B cells and T cells on the other side.

35:14.700 --> 35:19.700
And I should have said that up T cells and B cells on the other side of antigen presentation.

35:19.700 --> 35:24.700
So maybe that probably wasn't a very good explanation because it's no visuals here.

35:24.700 --> 35:36.700
But I think I understand this in a way that I think it makes sense that this works.

35:36.700 --> 35:41.700
I think it's a question of what can these antibodies be used for besides curing cancer.

35:41.700 --> 35:44.700
Like how well they neutralize the virus.

35:44.700 --> 35:51.700
We can see that even across donors, some factors jump out as being hotspots for potential neutralizers.

35:51.700 --> 35:57.700
So, for example, in each panel, there's this cluster of potent red antibodies in the lower left quadrant.

35:57.700 --> 35:59.700
I'll see if I can turn to Mike.

35:59.700 --> 36:02.700
So we don't even know what they stimulated this to, right?

36:02.700 --> 36:05.700
We don't know what the antigen was in this study.

36:05.700 --> 36:08.700
So it's really a vague description of what's going on.

36:08.700 --> 36:17.700
And has somebody said earlier, it's a bit like hypnosis, it's a bit, it's a bit salesy, it's a bit.

36:17.700 --> 36:22.700
It's a bit snake oily in that sense.

36:22.700 --> 36:25.700
This cluster here is what I'm talking about.

36:25.700 --> 36:27.700
It shows up in several of these panels.

36:27.700 --> 36:30.700
Good evening from Pittsburgh, Hong Kong.

36:30.700 --> 36:32.700
So this is exactly what we want to see.

36:32.700 --> 36:40.700
That our sequence embedding vectors are encoding useful information about the properties of the antibodies.

36:40.700 --> 36:46.700
So once we've identified antibodies of interest, first via imaging and then by looking at their sequences,

36:46.700 --> 36:50.700
we express and characterize them using several different assays.

36:50.700 --> 36:55.700
The first thing we want to do is validate that the antibodies bind as we actually expect they would.

36:55.700 --> 37:00.700
And then we have many other assays to measure things like chemical properties, their stability, their affinity,

37:00.700 --> 37:04.700
and the way they bind to our particular antigen.

37:04.700 --> 37:07.700
All this characterization data feeds our machine learning algorithms

37:07.700 --> 37:13.700
and can inform both upstream antibody detection and selection and also downstream protein engineering.

37:13.700 --> 37:18.700
And we continually accumulate more and more of this data as we run these assays.

37:18.700 --> 37:20.700
So make sure you heard that, right?

37:20.700 --> 37:23.700
He just told you a very special thing there.

37:23.700 --> 37:26.700
It's very similar to what EpiVac said.

37:26.700 --> 37:37.700
Annie DeGroote said, excuse me, that they can find those immunogenic epitopes.

37:37.700 --> 37:42.700
But then more importantly, the obvious dual use of that is to insert them into proteins,

37:42.700 --> 37:44.700
to make proteins more immunogenic.

37:44.700 --> 37:46.700
And he basically just said the same thing.

37:46.700 --> 37:48.700
You can use that.

37:48.700 --> 37:57.700
This information of these epitopes and the way they're encoded in antibodies also in protein design.

37:57.700 --> 38:05.700
And so biologics, proteins, protein design, I can't think of very many other things that you would want to design

38:05.700 --> 38:08.700
from the perspective of a vaccine other than immunogenicity.

38:08.700 --> 38:13.700
So controlling where that immunogenicity is on the protein

38:13.700 --> 38:18.700
and where the immune system focuses would be a pretty spectacular art form.

38:18.700 --> 38:24.700
And I think it's very possible that that's the story that's being manipulated here,

38:24.700 --> 38:30.700
that what they did was they told us a story about a sequence they found embedded in that sequence

38:30.700 --> 38:33.700
was a pre-designed immunogen.

38:33.700 --> 38:38.700
Whether or not that immunogen ever went in circulation I think is beyond a doubt not likely.

38:38.700 --> 38:45.700
But whether that immunogen was released as a transfection, as an infectious clone,

38:45.700 --> 38:52.700
with limited spread or with a large quantity over some of these places, that's very possible.

38:52.700 --> 38:58.700
And if it had an immunogenic protein on it, of course that would have generated a lot of the disease states

38:58.700 --> 39:02.700
that were necessary in order to start that panic in those different geographic regions.

39:03.700 --> 39:11.700
This technology to me has so much dual use, and the way that they took us so long to find this,

39:11.700 --> 39:17.700
and once we've found all of this, the way that the work of Mark Gulak and others,

39:17.700 --> 39:28.700
but for me, mainly Mark, has tied in some of the key players of the dissidents into this story.

39:28.700 --> 39:37.700
And one of those important ones, of course, is Robert Malone.

39:37.700 --> 39:42.700
We're both upstream antibody detection and selection, and also downstream protein engineering.

39:42.700 --> 39:47.700
And we continually accumulate more and more of this data as we run these assays,

39:47.700 --> 39:56.700
which means we can feed it back into our ML models and create a virtuous cycle to keep improving them.

39:57.700 --> 40:00.700
So at this point, I've described how we accumulate sort of a mountain.

40:00.700 --> 40:05.700
One of the things that you can imagine, which I just,

40:05.700 --> 40:12.700
I get frustrated when I think this kind of thing because I'm hesitant to say it because of the no virus people.

40:12.700 --> 40:20.700
But if we were to entertain the possibility that coronaviruses are a real RNA shadow in nature,

40:20.700 --> 40:23.700
that occasionally makes people sick all around the world,

40:24.700 --> 40:29.700
and that there's some homology between the yearly waves of coronavirus

40:29.700 --> 40:33.700
because the homology is around the most conserved proteins,

40:33.700 --> 40:36.700
including the RNA-dependent RNA polymerase.

40:36.700 --> 40:42.700
But then they said that there was a novel one, again with this designer immunogen on it,

40:42.700 --> 40:45.700
so that when they transfected people with it,

40:45.700 --> 40:51.700
a naturally occurring pathogen would become dangerous.

40:51.700 --> 40:56.700
And if you wanted to get rid of a large portion of a useless population,

40:56.700 --> 41:02.700
you might say everybody that takes this shot will just be more vulnerable to the immune interaction

41:02.700 --> 41:07.700
with these previously harmless coronaviruses,

41:07.700 --> 41:11.700
which again, according to their mythology, have so much overlap in their proteins

41:11.700 --> 41:15.700
that according to anti-dechrot, or dechrot,

41:15.700 --> 41:20.700
the protein would actually be tolerogenic.

41:20.700 --> 41:27.700
And so that's no good if all of the coronavirus proteins are mimicking human epitopes

41:27.700 --> 41:31.700
so that they generate tolerance,

41:31.700 --> 41:35.700
then you have a real issue because then how do you make a vaccine to that?

41:35.700 --> 41:36.700
You don't.

41:36.700 --> 41:39.700
But you could make a vaccine to it if you designed the spike protein

41:39.700 --> 41:42.700
that you intended to use as a vaccine to be immunogenic

41:42.700 --> 41:47.700
and then lied about it being present in a novel lab league.

41:47.700 --> 41:50.700
And I know I'm going out here on a limb here, but what?

41:50.700 --> 41:54.700
I mean, otherwise I'm repeating myself over and over again.

41:54.700 --> 42:00.700
I think this is the sort of intellectual space you have to explore at this stage,

42:00.700 --> 42:08.700
because if this is the front-facing, public, non-confidential presentation that this guy gives,

42:08.700 --> 42:16.700
then I can anticipate what the behind-the-scenes DARPA DOD presentation is.

42:16.700 --> 42:18.700
That's for sure.

42:18.700 --> 42:21.700
And I can certainly imagine what Dr. Giordano would take with this

42:21.700 --> 42:25.700
and say, well, what we can do with this is a lot.

42:25.700 --> 42:30.700
So to me, that's pretty easy to see.

42:30.700 --> 42:33.700
Anyway, I'm going to let this go for a little while longer,

42:33.700 --> 42:36.700
because Mark might not be ready till 8.30,

42:36.700 --> 42:38.700
so I'm going to go a little while longer.

42:38.700 --> 42:42.700
If he doesn't ping me back, I'll just end in a few minutes anyway.

42:42.700 --> 42:45.700
Data, characterizing the antibodies we find.

42:45.700 --> 42:49.700
But it's likely that there's still too many antibodies to bring to a clinical trial.

42:49.700 --> 42:52.700
So we have to further refine the list.

42:52.700 --> 42:55.700
So we've developed this internal tool, we call it Celium,

42:55.700 --> 42:59.700
to help our scientists make sense of this very high-dimensional data set.

42:59.700 --> 43:01.700
And so that's really the trick, right?

43:01.700 --> 43:03.700
That's what they're selling.

43:03.700 --> 43:08.700
They're doing some pretty basic microfluidic sorting of B cells.

43:08.700 --> 43:12.700
And the actual selling thing, I think, honestly,

43:12.700 --> 43:16.700
after they have all this high-throughput screening of B cells done.

43:16.700 --> 43:18.700
And the main thing that they're selling,

43:18.700 --> 43:22.700
the main thing that they're going to leverage are AI, machine learning,

43:22.700 --> 43:26.700
software stacks that contain patented code,

43:26.700 --> 43:30.700
or patented subroutines, or patented search parameters, yada, yada,

43:30.700 --> 43:34.700
yada, whatever it is, trademark names.

43:34.700 --> 43:39.700
And that's how they layer on the value that they provide to their clients.

43:39.700 --> 43:43.700
That's how they layer on the value that they draw for the royalties

43:43.700 --> 43:46.700
that they're going to charge when they return the data back to those people,

43:46.700 --> 43:49.700
and then they use it to make products and they owe them royalties.

43:49.700 --> 43:52.700
That's the whole model.

43:52.700 --> 43:57.700
And so companies that bring their old monoclonals to them

43:57.700 --> 44:02.700
can launder them through here and generate new monoclonals

44:02.700 --> 44:06.700
to the same target that might be better and then have launder them through

44:06.700 --> 44:09.700
or produce them with this software.

44:09.700 --> 44:17.700
So it's a patentable process that enables the finding of monoclonal antibodies.

44:17.700 --> 44:22.700
And so you just bypassed the antibody patent paradox completely,

44:22.700 --> 44:25.700
and then you did it in Canada, no matter, even better.

44:25.700 --> 44:29.700
See, I think it's a really interesting story that more people,

44:29.700 --> 44:34.700
with more knowledge than I, need to latch onto it,

44:34.700 --> 44:37.700
start to investigate how plausible this is.

44:37.700 --> 44:40.700
And meanwhile, they have Dr. Giordano running around and telling everybody

44:40.700 --> 44:45.700
about all this other stuff that's actually still 50 years on the horizon

44:45.700 --> 44:48.700
if that, where all of this stuff is real and right now

44:48.700 --> 44:52.700
and totally relevant for the pandemic narrative.

44:52.700 --> 44:54.700
So I don't know what to say.

44:55.700 --> 44:58.700
It overlays information derived from the protein sequences,

44:58.700 --> 45:01.700
alongside measurements collected at the wet bench.

45:01.700 --> 45:05.700
So it gives scientists a way to tease apart and understand these relationships.

45:05.700 --> 45:09.700
And we're constantly integrating more powerful ML-backed analysis tools

45:09.700 --> 45:13.700
into Selium to help our scientists understand the antibody landscape

45:13.700 --> 45:15.700
and make better decisions.

45:19.700 --> 45:23.700
So up to now, I've told you about our single cell screening platform

45:23.700 --> 45:27.700
and sort of to recap that, it all starts with that immune response.

45:27.700 --> 45:30.700
And from that immune sample, we screen single B cells in isolation.

45:30.700 --> 45:33.700
We're looking for interesting binding signatures.

45:33.700 --> 45:37.700
And then we recover those B cells and analyze their amino acid sequence.

45:37.700 --> 45:40.700
And finally, we express the best candidates, characterize their properties,

45:40.700 --> 45:42.700
and validate their binding profile.

45:42.700 --> 45:46.700
But because the immune system can produce such a vast diversity of antibodies,

45:46.700 --> 45:52.700
screening even 500,000 cells isn't enough for a complete picture of the immune response.

45:54.700 --> 45:57.700
So in parallel with this single cell screening process,

45:57.700 --> 46:01.700
we can also sequence an even larger sample of the immune cells using a process

46:01.700 --> 46:03.700
called repertoire sequencing.

46:03.700 --> 46:08.700
And this gives us even more sequence data about what the immune system is doing

46:08.700 --> 46:10.700
and the sequences that are present in the sample.

46:10.700 --> 46:12.700
But it comes at a cost.

46:12.700 --> 46:15.700
We get much less information about each individual antibody.

46:15.700 --> 46:21.700
So the single cell process that's depicted in the top row and that I described in detail

46:21.700 --> 46:25.700
tells us exactly which heavy and light chains were paired together

46:25.700 --> 46:27.700
to form a particular antibody.

46:27.700 --> 46:31.700
But the repertoire sequencing process mixes all the heavy and light chains together

46:31.700 --> 46:34.700
so the sequences come to us unpaired.

46:34.700 --> 46:39.700
And it's also harder to look for a very specific binding profile using this technology.

46:41.700 --> 46:45.700
To deal with this, we use bioinformatics and machine learning techniques

46:45.700 --> 46:49.700
inspired by natural language processing to learn features from our single cell antibodies

46:49.700 --> 46:52.700
and find related sequences from the repertoire data.

46:52.700 --> 46:58.700
So here on the left, you can see an inferred lineage relationship between a single cell lead.

46:58.700 --> 47:00.700
This is our antibody of interest.

47:00.700 --> 47:01.700
It's highlighted in green.

47:01.700 --> 47:03.700
And then it's germline parent, which is above it.

47:03.700 --> 47:08.700
And that's a putative ancestor of this cell.

47:08.700 --> 47:13.700
And then some of its single cell descendants that also showed up in the assay,

47:13.700 --> 47:16.700
the single cell screening assay.

47:17.700 --> 47:20.700
But now on the right, what you see is a much deeper lineage tree,

47:20.700 --> 47:24.700
and it includes heavy chains obtained from repertoire sequencing.

47:24.700 --> 47:28.700
This same single cell heavy chain is still shown in green on the right-hand tree,

47:28.700 --> 47:33.700
but now it fits into a much richer set of the heavy chains obtained from the repertoire sequencing.

47:33.700 --> 47:36.700
And this helps us understand how the single cell hits we've isolated

47:36.700 --> 47:39.700
are fitting into the overall immune response,

47:39.700 --> 47:44.700
and to potentially identify even better candidate sequences.

47:45.700 --> 47:47.700
I can freely admit I'm not really following.

47:47.700 --> 47:49.700
So stepping back to the bigger picture of how ML fits into eCelera,

47:49.700 --> 47:53.700
I see it as much less of an end product than as a research and problem-solving tool.

47:53.700 --> 47:57.700
So here machine learning exists to support a never-ending stream

47:57.700 --> 48:02.700
of partner discovery projects and campaigns across the entire scientific pipeline.

48:02.700 --> 48:08.700
We're driving scientific research, so there's often a new twist on every project.

48:08.700 --> 48:12.700
One problem we definitely don't have is running out of our validation and test data.

48:12.700 --> 48:17.700
But what we do constantly worry about is whether our models will generalize into their next deployment.

48:17.700 --> 48:22.700
So we have to make fluid transitions between exploratory data analysis and machine learning

48:22.700 --> 48:25.700
to understand where we are and how we're doing.

48:25.700 --> 48:29.700
Our goal isn't just to automate specific processes or fine-tune models

48:29.700 --> 48:32.700
to improve metrics on canned pre-existing datasets.

48:32.700 --> 48:35.700
eCelera controls the entire discovery pipeline from immunization strategies,

48:35.700 --> 48:38.700
cell screening protocols, imaging equipment, mechatronics,

48:38.700 --> 48:41.700
protein expression, assay development, all of this.

48:41.700 --> 48:44.700
So we have a say in how the datasets are generated, what gets measured,

48:44.700 --> 48:47.700
and which processes can be improved.

48:47.700 --> 48:52.700
Something else about our team that's somewhat unique is that we operate across many different data scales.

48:52.700 --> 48:56.700
Some parts of our pipeline generate terabytes of data relatively cheaply,

48:56.700 --> 49:01.700
whereas other assays are so costly or slow that we only get a handful of data points to work with.

49:01.700 --> 49:08.700
So we have to be creative and make sure we can gracefully jump to the right scale.

49:08.700 --> 49:13.700
I want to spend just a minute recounting a story that everyone at eCelera is very proud to be part of.

49:13.700 --> 49:18.700
So in early 2020, as you all know, the COVID pandemic was starting to spread throughout the world,

49:18.700 --> 49:21.700
and this put our technology through a trial by fire.

49:21.700 --> 49:28.700
Within three days of receiving a sample from one of the first US patrons who had recovered from COVID-19,

49:28.700 --> 49:31.700
we had screened 5.8 million cells.

49:31.700 --> 49:36.700
And 23 days later, we had selected just 24 leads from those 5.8 million.

49:36.700 --> 49:43.700
And within 90 days, through a partnership with Eli Lilly, one of those antibodies was already in human clinical trials.

49:43.700 --> 49:49.700
Bam Lanovab has gone on to become the first potential treatment developed specifically for COVID-19,

49:49.700 --> 49:55.700
and it's approved in more than 10 countries and approximately 400,000 patients have been treated.

49:55.700 --> 49:59.700
It was a lot of work, and we're very proud of this accomplishment.

50:00.700 --> 50:09.700
I'm wondering if you were using just ML based models, or do you use a mixture of mechanistic or biochemical models?

50:09.700 --> 50:14.700
We have a huge range of models, and this is exactly the kind of question that we're exploring at eCelera.

50:14.700 --> 50:20.700
I can't talk about all the different kinds of models we're using, but we are trying to find what,

50:20.700 --> 50:25.700
and that is what our daily life at eCelera is like, is solving these kinds of research problems.

50:26.700 --> 50:33.700
So that's pretty extraordinary, because that's really the kind of bamboozlement that leads to a pandemic.

50:33.700 --> 50:35.700
It's really how they did it.

50:35.700 --> 50:42.700
It's how they sell stock, it's how they sell these bad ideas on the road.

50:42.700 --> 50:49.700
It's making promises that they can't keep, and as they said, forward-looking statements may not be accurate.

50:50.700 --> 50:56.700
And starting to talk with that, and then talking about this biology as if it's all but done,

50:56.700 --> 51:00.700
I have no doubt that there are uses for monoclonal antibodies.

51:00.700 --> 51:09.700
I have no doubt that there are ways that the immune system can be used in this way to produce therapies for things like cancer and other conditions,

51:09.700 --> 51:16.700
but I do not think we are anywhere near the stage where we can usefully augment the memory of the immune system.

51:17.700 --> 51:24.700
And this talk of doing that was, he started it in the beginning, but he didn't say a word about it afterward.

51:24.700 --> 51:38.700
And so I hope what you can see here really plain as day is that this is just a very multifaceted way of doing an end-around on the antibody-patent paradox,

51:39.700 --> 51:50.700
and the idea that antibodies, monoclonal antibodies in particular, have been patented in a way in the past that they can no longer be patented.

51:50.700 --> 52:00.700
And the entire sort of intellectual property space that is monoclonal antibodies has changed drastically in the past few years because of a recent Supreme Court ruling.

52:01.700 --> 52:14.700
And it's possible that that sort of momentum was known for a while because of previous rulings, including the one that Brett Weinstein's father set in on,

52:14.700 --> 52:18.700
that Mark has reported a number of times, Santa Corps versus somebody else.

52:18.700 --> 52:30.700
And so that's really a point where we are, where we can see a play, a play for one of the only things that does work for certain things.

52:30.700 --> 52:45.700
I'm still curious as to whether monoclonal antibodies really made any difference for people with COVID or whether the antibodies themselves can just ramp up an immune response in a way that is effectively useful for anybody that gets monoclonal antibodies.

52:45.700 --> 52:52.700
Because if there's no meaningful amount of virus in your blood, then monoclonal antibodies shouldn't in theory do anything.

52:52.700 --> 52:57.700
And so I'm still trying to figure out what's going on here too.

52:57.700 --> 53:04.700
I don't have all of the things working in my head, but I know for sure I heard Tad Hughes say that they have so many models and we can't talk about them.

53:04.700 --> 53:13.700
And the whole point of having models that aren't validated or having ideas that are patented first and validated later, this is all business.

53:13.700 --> 53:15.700
It's very different than biology.

53:15.700 --> 53:30.700
And a lot of the pandemic biology is based on business biology, like this talk, and not about real proven understood, you know, grassroots, not grassroots, but foundational immunology.

53:31.700 --> 53:36.700
So I think that's what we're really trying to work on here now.

53:36.700 --> 53:53.700
And I'm trying to build a new slide set that will more adequately describe the absolute state of where we are with understanding this and the kind of possibilities that are there from the worst case scenarios to the most ridiculous ones.

53:54.700 --> 54:02.700
But I think in general, it's a very good place to start for everybody from the perspective of if you've never heard this before.

54:02.700 --> 54:22.700
Or if you're trying to tell somebody something that they didn't expect to hear, you should start with the idea that algorithms and the NIH and the NIAID and DOD and everybody involved in the world public health space has had an impetus

54:22.700 --> 54:33.700
and a notion in their head that pandemics have potential and that potential can be found in mother nature.

54:33.700 --> 54:38.700
If you go looking for it, it can be extracted from mother nature using cell culture.

54:38.700 --> 54:50.700
It can be extracted from mother nature using animal passage and it can even be kind of woven together using genetic techniques to make viruses that maybe mother nature herself never would have come up with.

54:50.700 --> 55:10.700
And this combination of mythologies is being used to convince everybody that there's so much danger from a pandemic and that digital IDs make sense that mandatory vaccinations make sense and that health freedom is really a thing of the past.

55:11.700 --> 55:18.700
And they most importantly want our children to believe that. So that's why we really need to be outspoken and we need to act.

55:18.700 --> 55:27.700
We can't just wait for everybody else to fix. We can't wait for people to wake up. Everyone needs to wake up and you need to demonstrate that you're awake.

55:27.700 --> 55:36.700
And so you need to see this picture for what it is. The all-cause mortality is in light blue and that is the graph that they very rarely showed us.

55:36.700 --> 55:47.700
You can see this graph sometimes in the UK but in America they never showed it to us and more importantly they never showed us the change in the deaths due to pneumonia.

55:48.700 --> 55:57.700
And if you just look at this graph and see the middle color blue here on here on here on here the same number of people are dying of pneumonia.

55:57.700 --> 56:08.700
That's this line right here. And that same number of people dying of pneumonia also represents a huge number of people that are having pneumonia even going to the hospital with it and not dying.

56:09.700 --> 56:22.700
And so there is a huge potential for an increase in all-cause mortality due to pneumonia alone if you just change the way it's treated or prevent its treatment.

56:22.700 --> 56:32.700
And for quite some time now I have been arguing that that is one of the primary sources of this limited excess deaths.

56:32.700 --> 56:45.700
That signal was confirmed by Denny Rancor last year and the year before and shown that the all-cause mortality is not correlated with a spreading pathogen.

56:45.700 --> 56:51.700
It stops at borders and more importantly it's correlated with the income of the location.

56:51.700 --> 56:58.700
It can be also correlated with a geographic location. It doesn't cross state borders.

56:58.700 --> 57:02.700
It doesn't cross country borders. It's pathetic. It's the protocols.

57:02.700 --> 57:15.700
And the way that they did it was they changed those protocols with respect to pneumonia by using the implication of a novel virus and the idea that you could test for it if you had a test.

57:15.700 --> 57:22.700
And if you remember in 2020 we didn't have a test in America until March or April, maybe even April or May.

57:22.700 --> 57:29.700
And so a lot of the COVID that was actually in New York City was presumed COVID. Not tested COVID but presumed COVID.

57:29.700 --> 57:35.700
And so that's also a big problem ladies and gentlemen. A huge problem.

57:35.700 --> 57:50.700
And I think that the biology of RNA and the biology of RNA replication as it is drawn best by them means that a natural coronavirus swarm can sustain a pandemic and that's why we have never had one.

57:50.700 --> 58:00.700
That's why SARS burnt out. That's why mayors burnt out even when you make a clone and you release it in China even when you make a clone and you release it in the Middle East.

58:00.700 --> 58:05.700
It's not going to go anywhere. And the clone is a best case scenario.

58:05.700 --> 58:19.700
And that's something that I haven't really made very clear but a clone is the best case scenario for transmission because a clone is millions of copies of the RNA in complete form.

58:20.700 --> 58:31.700
Whereas any other form of the natural virus is a swarm, mostly in competent particles, hundreds of thousands of times more subgenomic RNAs than full genomes.

58:31.700 --> 58:37.700
And so if you don't have very full genomes and all of them are different and many of them are replication and competent.

58:37.700 --> 58:48.700
And then it's obvious that a natural swarm can't sustain a pandemic but if you released a clone in several places it would appear as though a pandemic was underway.

58:48.700 --> 59:00.700
And that high homology between all of those sequences would be your excuse and also would be the magic that would fool all of those molecular biologists around the world to believe there really was an ongoing pandemic.

59:00.700 --> 59:13.700
Because heretofore the identical RNA sequence of a virus has never been found simultaneously in multiple places in the world at the same time. Never.

59:14.700 --> 59:27.700
And so they're using this mythology to invert the rights and individual sovereignty of your children into basic granted permissions.

59:27.700 --> 59:31.700
They want you to carry an ID and ask for permission to go everywhere.

59:31.700 --> 59:39.700
And unfortunately we have been slow rolled into this narrative when a lot of these people could have woken up in 2020.

59:39.700 --> 59:43.700
Mostly didn't. Here at Fund in Bosch was nowhere to be found.

59:43.700 --> 59:45.700
Robert Malone was nowhere to be found.

59:45.700 --> 59:52.700
Brett Weinstein was wearing a bandana and isolating all through 2020.

59:52.700 --> 59:59.700
And pushing a drastic lab leak scenario very early on.

59:59.700 --> 01:00:08.700
So all of these individuals that we've been trying to investigate are responsible for this illusion of consensus that we have to find out where this virus came from.

01:00:09.700 --> 01:00:14.700
And that the virus did most of it is one of the things that people are still insistent of.

01:00:14.700 --> 01:00:20.700
And that's where you can just see the liars right away when they emphasize the virus still in 2023.

01:00:20.700 --> 01:00:23.700
I mean, give it a rest already.

01:00:23.700 --> 01:00:31.700
Whatever bio weapon was released in 2020 is not represented here now in 2023. That's just absurd.

01:00:32.700 --> 01:00:40.700
And the idea that they're still making that argument in light of all of the biology that we brought out on Giggle Home. It's just sad, really.

01:00:40.700 --> 01:00:48.700
If there's anything going on in the brain, if there's anything going on with amyloidosis or anything like that, it's going to be from transfection.

01:00:48.700 --> 01:01:01.700
Because transfection is the use of synthetic modified RNA to express foreign proteins in your body, and they can't control where it goes, as Peter Cullis told us the other night.

01:01:01.700 --> 01:01:13.700
And so that is a much better, much finer recipe for all of these protein misregulations than a released coronavirus circulating the world is.

01:01:14.700 --> 01:01:26.700
And least of all, because the released coronavirus could never have the fidelity to sustain and keep a hold of all these magic epitopes that supposedly are going around the world. It's just not possible.

01:01:26.700 --> 01:01:32.700
And all these people who say that it is can't produce any papers that show that it is.

01:01:32.700 --> 01:01:39.700
The only papers that are available are papers that make the exact argument that I'm making, which is incompetent fractions.

01:01:39.700 --> 01:01:50.700
It is a genetic swarm, and these two things result in a biology that is different than what they have portrayed on television over the last three years.

01:01:50.700 --> 01:01:56.700
But in order for them to get away with this, they needed to change the way we thought about all of this stuff.

01:01:56.700 --> 01:02:00.700
And they needed to change the way we think about the cause of respiratory disease.

01:02:00.700 --> 01:02:06.700
And they needed to completely bamboozle with us about all cause mortality in America. They never said once.

01:02:06.700 --> 01:02:13.700
That between 55 and 65,000 Americans are expected to die every week.

01:02:13.700 --> 01:02:21.700
They never made any reference to all cause mortality of being close to 3 million per year.

01:02:21.700 --> 01:02:29.700
And so even as they were inverting and substituting and driving up opioid deaths and blaming it all on COVID,

01:02:29.700 --> 01:02:39.700
they never took the time to frame these numbers in the context of all cause mortality because their intention was to make it seem like there was a novel cause of death.

01:02:39.700 --> 01:02:42.700
Something went from zero to a hundred.

01:02:42.700 --> 01:02:58.700
When in reality they were just swapping things around and there wasn't any change except for in the places where they were managing to convince hospitals to change their protocols and doctors to change their protocols.

01:02:58.700 --> 01:03:05.700
And so they ventilated people to prevent spread and killed many. They used remdesivir, medazolam, and they didn't treat pneumonia.

01:03:05.700 --> 01:03:15.700
It's very simple. Three pills were removed from the drawer and everything else was an illusion of consensus.

01:03:15.700 --> 01:03:19.700
All these people agreeing can fool you just like an ash conformity experiment.

01:03:19.700 --> 01:03:25.700
They have fooled us into trying to figure out whether it was a laboratory or a bad cave virus zoonosis.

01:03:26.700 --> 01:03:34.700
In these scenarios, lockdowns and emergency use authorizations caused some excess deaths, but the millions were killed by a novel virus.

01:03:34.700 --> 01:03:39.700
In these scenarios we've defeated epidemics in the past with vaccination.

01:03:39.700 --> 01:03:43.700
In this scenario, novel coronaviruses can jump species in pandemic.

01:03:43.700 --> 01:03:47.700
In this scenario, PCR false positives are very rare.

01:03:47.700 --> 01:03:49.700
Asymptomatic spread is real.

01:03:49.700 --> 01:03:56.700
And we spend money studying viruses, including gain of function experiments that can generate pandemic potential.

01:03:56.700 --> 01:04:00.700
This is the TV scenario one through three.

01:04:00.700 --> 01:04:09.700
The real scenario I believe is that there was a conflated background signal and what that conflated background signal was is not necessarily my job to find out,

01:04:09.700 --> 01:04:17.700
but it could have been the background signal of coronaviruses or whatever they are in reality if their exosomes or whatever.

01:04:18.700 --> 01:04:30.700
And it could also be the combination of that background signal could also be the RNA signals that are present naturally in our nasal passages all the time.

01:04:30.700 --> 01:04:37.700
And so they could be human sequences, bacterial sequences, fungal sequences, whatever.

01:04:37.700 --> 01:04:45.700
And as long as there's sufficient overlap with the sequence that they're looking for, with their primers, the N protein or the RNA,

01:04:45.700 --> 01:04:50.700
dependent RNA polymerase or whatever they say they're using for their, it's ultimately down to the primers.

01:04:50.700 --> 01:04:52.700
It's not about what it's aimed at.

01:04:52.700 --> 01:05:04.700
If they choose primers which are not exquisitely specific for the target gene and overlap with many of the other background,

01:05:04.700 --> 01:05:12.700
RNA and DNA that's known to be present in your nose or in your saliva, then a false positive is not a false positive.

01:05:12.700 --> 01:05:19.700
You're getting a positive and an amplicon is being amplified, but it's just not the target amplicon that they said it was.

01:05:19.700 --> 01:05:24.700
And so now very quickly you roll out a bunch of these things like we did in America.

01:05:24.700 --> 01:05:32.700
There were 250 different products called PCR tests, 250 products.

01:05:32.700 --> 01:05:35.700
I mean, think about that.

01:05:35.700 --> 01:05:36.700
And so it doesn't really matter.

01:05:36.700 --> 01:05:37.700
And I've said this a long time.

01:05:37.700 --> 01:05:41.700
It doesn't really matter whether it was a leak, a lease, a release.

01:05:41.700 --> 01:05:42.700
It's all lies.

01:05:42.700 --> 01:05:44.700
They've been lying about it.

01:05:44.700 --> 01:05:51.700
And once you realize that they're lying about it and you realize that the protocols were murder and transfection is in medicine,

01:05:51.700 --> 01:05:57.700
then how you divvy up this, the cause and effect here doesn't really matter if you know the protocols were murder.

01:05:57.700 --> 01:06:03.700
There's hardly anything left for a particular ravenous virus to do anything with.

01:06:03.700 --> 01:06:09.700
There's no reason to even have it in the story, quite honestly, because there was so much terrible things,

01:06:09.700 --> 01:06:13.700
so many terrible things done with regards to protocols and treatments.

01:06:13.700 --> 01:06:18.700
And so infectious clones released in Iran and Wuhan, sure, why not?

01:06:18.700 --> 01:06:22.700
A transfection agent released somewhere in Iran and Wuhan, sure, why not?

01:06:22.700 --> 01:06:23.700
Whatever you want to call it.

01:06:23.700 --> 01:06:25.700
But this isn't the right answer.

01:06:25.700 --> 01:06:27.700
And this probably is.

01:06:27.700 --> 01:06:29.700
That's really, I really think that's where we are.

01:06:29.700 --> 01:06:32.700
And again, I will keep saying this over and over again.

01:06:32.700 --> 01:06:34.700
Let's pay attention to what people talk about.

01:06:34.700 --> 01:06:43.700
If they're not all going back to the beginning and pointing out the depth and the breadth of the fraud that took place in

01:06:43.700 --> 01:06:52.700
2020 and early 2021 to assure that warp speed would result in lots of people being vaccinated right after

01:06:52.700 --> 01:06:58.700
Joe Biden was elected, then these people aren't being honest with you.

01:06:58.700 --> 01:07:02.700
If they're not looking into Jessica Hockett's data, they're not being honest with you.

01:07:02.700 --> 01:07:08.700
If they're not looking into Denny Rancour's data, they're not being honest with you.

01:07:08.700 --> 01:07:16.700
If they're not aware of the fact that the same thing that Jessica Hockett found in New York was also found in Italy,

01:07:16.700 --> 01:07:18.700
they're not being honest with you.

01:07:18.700 --> 01:07:24.700
If you haven't seen that what Jessica Hockett found in New York is almost so basically found in Chicago,

01:07:24.700 --> 01:07:26.700
then people aren't honest with you.

01:07:26.700 --> 01:07:36.700
If people aren't sharing the fact that the antibody patent paradox case actually occurred during the pandemic

01:07:36.700 --> 01:07:43.700
and that all of these companies that seem to be bypassing or are doing an end-around on the antibody patent paradox

01:07:43.700 --> 01:07:49.700
and that intellectual property space all have a really weird red thread of Robert Malone,

01:07:49.700 --> 01:07:56.700
including a lot of the Canadian companies, including the technology and interacting with some of these companies

01:07:56.700 --> 01:08:00.700
for many, many years, people aren't being honest with you.

01:08:00.700 --> 01:08:07.700
They're talking about something else, talking about something else and asking the wrong questions,

01:08:07.700 --> 01:08:12.700
tricking you into asking the wrong questions, and there aren't too many more questions to ask really.

01:08:13.700 --> 01:08:15.700
Except why haven't we stopped this train?

01:08:18.700 --> 01:08:24.700
And the reason why is because they want our data, yes they do, they would like us to put sensors in our skin

01:08:24.700 --> 01:08:27.700
and all this other stuff, and more importantly they want our kids to do it.

01:08:27.700 --> 01:08:31.700
Remember, I think this is a long game, it's for all the marbles and I don't think they need us to do it.

01:08:31.700 --> 01:08:34.700
I don't even know if they need us to be around before they do it.

01:08:34.700 --> 01:08:40.700
It could be that they're really aiming at my son's kids, your kids, kids.

01:08:41.700 --> 01:08:47.700
And it's that long of a plan, like a 50-year plan, it's not a 10-year plan or a 20-year plan.

01:08:47.700 --> 01:08:50.700
Oh, it's 8.30, I got to get out of here.

01:08:50.700 --> 01:08:53.700
Ladies and gentlemen, intramuscular injection of any combination of substances

01:08:53.700 --> 01:08:55.700
with the intent of augmenting the immune system is dumb.

01:08:55.700 --> 01:08:57.700
Transfection is not immunization.

01:08:57.700 --> 01:09:00.700
Please stop all transfections in humans.

01:09:00.700 --> 01:09:06.700
And I apologize for kind of an impromptu improvised stream here.

01:09:06.700 --> 01:09:10.700
I know I'm usually a little more organized, but I wanted to do something short,

01:09:10.700 --> 01:09:14.700
and then Mark gave me an extra half an hour, so I went a little longer than I should have.

01:09:14.700 --> 01:09:16.700
Make sure you follow Mark, make sure you go find him.

01:09:16.700 --> 01:09:20.700
He's on YouTube right now, I believe, so I might go look for that.

01:09:20.700 --> 01:09:24.700
I might go look for that link and drop it in right here.

01:09:24.700 --> 01:09:26.700
See how I can find it.

01:09:26.700 --> 01:09:33.700
There we go.

01:09:33.700 --> 01:09:41.700
There it is.

01:09:41.700 --> 01:09:45.700
I'll be over here in a little while.

01:09:45.700 --> 01:09:53.700
Thanks, guys, for joining me.

01:09:53.700 --> 01:09:58.700
Yep, thanks very much for joining me, everybody, and I will see you again tomorrow for another show.

01:09:58.700 --> 01:10:03.700
Hopefully my voice will still sound okay.

01:10:16.700 --> 01:10:17.700
What was that?

01:10:17.700 --> 01:10:20.700
Oh, that's him, he's starting.

01:10:20.700 --> 01:10:24.700
Oh, see you guys.

01:10:24.700 --> 01:10:33.700
Take care.

01:10:33.700 --> 01:10:36.700
Make sure you go over to Mark.

01:10:36.700 --> 01:10:40.700
Go over to Mark's show, say hi to him for me.

01:10:40.700 --> 01:10:44.700
Let's see if we can get his viewers up to 100 before he goes to sleep.

01:10:44.700 --> 01:10:49.700
Thanks for joining me, guys, I'll see you tomorrow.