U.S. Transhumanist Party Virtual Enlightenment Salon - Rafi Qumsieh - KGRA Digital Broadcasting Discovery

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Greetings and welcome to the United StatesTranshumanist Party Virtual Enlightenment Salon. My name
is Jannati Stoliero the second and Iam the Chairman of the US Transhumanist Party.
Here we hold conversations with some ofthe world's leading thinkers in longevity,
science, technology, philosophy and politics. Like the philosophers of the Age of

(00:22):
Enlightenment, we aim to connect everyfield of human endeavor and arrive at new
insights to achieve longer lives, greaterrationality, and the progress of our civilization.
Greetings, ladies and gentlemen, andwelcome to our US Transhumanist Party Virtual
Enlightenment Salon of a sober twenty secondtwenty twenty three. I am pleased to

(00:47):
offer you today a discussion of howmachine learning can be harnessed to advanced drug
discovery, particularly in the field oflongevity, where of course, it is
of paramount interest to us to personallyreach longevity escape velocity. Yet there are

(01:07):
so many options to consider and investigate, So how can this process be accelerated?
How can we determine which targets areactually worthwhile to investigate and focus on?
And this is where artificial intelligence andmachine learning can become extremely helpful.

(01:29):
So joining us today is our distinguishedpanel of US Transhumanist Party officers and members,
including our Director of Visual Art ArtRamon Garcia, our Director of Longevity
Outreach Ben boalweg Are, twenty twentyfour US Vice presidential candidate and Director of
Citizen and Community Science Daniel Tweed,our member Alan Crawley, and our special

(01:55):
guest today is doctor Yvonne de Weber. He is a PhD as scientist and
entrepreneur behind pioneering concepts in aging,genetics and stem cells. He turned to
pharmaceutical bioinformatics amid an explosion in biomedicaldata. His quest for faster drug development
led to the creation of core TeX'sDiscovery, his company which uses machine learning

(02:21):
to accelerate the search for new medicines, secure disease, and expand human longevity.
So thank you, yvon and welcome. Please tell us a bit more
about yourself and the work of Cortech'sDiscovery, and I know you have quite
an intriguing presentation in store for us. Okay, great, So guys,

(02:46):
maybe ever going to start by thebeginning. I think one of my first
memories was my parents telling me thatpeople had died and some of their friends.
So I was wondering and I wasasking them questions like do we have
to die? Everyone has to die, but not us, But they said,

(03:06):
no, we have to die aswell. And I think that that
was a big It was really mybig, biggest, strongest memory that I
might remember. So I started tothink about I don't know, philosophy,
religion, but nothing didn't really fitlogics, so I decided to find a

(03:27):
way to live longer. Back inthe time, at around ten, I
met my current business partner for Cortex, and I told him I wanted to
live forever, forever, and hereplied that he would rather replace human kind
by Roberts. So fifteen years after, I got my PhD in the field
of edging genetics, and in themeantimes, I really changed my plans because

(03:51):
at the beginning, I thought,probably I will need to find a way
to get frozen in a proper wayand then have a second chance later on.
But when I started to investigate themechanisms of aging, I was,
for instance, talking to one ofmy professor in biology and I asked him

(04:11):
if what would happen if we tweaka little bit at that mechanism. In
that direction, I think we wouldlive longer. And the gay was thinking
a little bit a few seconds andtold be, yeah, you might be
right, but what for? Sothen I realized that there is a problem
that's maybe linked to many parameters thatpeople they have accepted things, but we

(04:35):
are arriving at an age where manythings start to be possible. And I
completely gave up my curl conservation thingand I decided to work really towards like
extending lifespan and to the stage wherewe would be able to repair ourselves as
fast as we decay. So Ithink we can or the presentation. So

(05:02):
I was looking at data because Ihad some targets in mind for longevity that
I kind of was thinking about duringmy PhD. Have done some work with
this, and the problem is thatyou might have some targets, but it
takes a tremendous amount of time andmoney to develop a drug at a very

(05:25):
low success rate. So I startedto do a few businesses in the meantime,
thinking about how fast we can dosomething actionable. So before I was
completely against everything that was natural orwhatever, but when I realized that some
natural compounds are extending the expression increasingthe expression of some genes. I started

(05:50):
to be involved in in epigenetic nutrascity cars or even costmesotic cars. But
I was just like on the sidebecause I was waiting that the technology get
cheaper so I could start achieving mygoals by starting their research. But then
still things were very slow. SoI decided to investigate chemin forumatics, and

(06:13):
then I realized there was a biggap in machine learning. So I contacted
my old friend who was back inthe time professor in quantum information, and
I convinced him to retire and westarted Cortex. Now we are a team
of five mostly machine learning, alittle bit of quantum chemistry, and we're

(06:38):
working on two technologies. The firstone it's like an advanced version of what
people called QSR like quantitative activity concuntitivestructure activity relationship. It's basically learning from
experiment. The input is americle,the output is a property. And we're
working on a second technology that iscalled ducking, where we look really at

(07:04):
the three D interactions between compounds.The first technology is already available and work,
so I'm going to talk about thatone first. So we gathered all
the data that we get mostly fromBubken and Campbell. It's the largest representitor

(07:25):
is of high input screening data andalso from publication looking at pharmacological data about
compounds. So we have about twohundred and sixty million data points on this
database, and we build some neuralnetworks where the input is compound and the

(07:47):
output is a property. These aredeep graph neural nets, and we train
these neural nets to predict every propertythat we have in our database. At
the same time. It's about twothousand properties. But some properties can be
anything like, for instance, thenumber of cells that die, an activation

(08:11):
of a targets, an inhibition ofa pathway. We just optimize actually the
compound for a number. So whatwe are doing in terms of machine learning
is like it's basic, it's tendards. So we're using let's say eighty percent

(08:31):
of a data set where do inputis a compound and the output is a
property, and the neural net ismaking a correlation between the molecules structure and
the property, and then we helpdata. We're just plugging the twenty percent
of the compounds and we make acomparison between the predictions that we do with

(08:52):
the twenty percent of compounds and theactual reality, and that's what you can
see in the red and blue graph. So this is the testing sets or
the active compounds are the red compoundsand the inactive compounds are the blue compounds.
So here is good because the activecompounds we give them a high probability

(09:16):
of activity and the inactive compounds alow pity of activity. And the way
we actually quantify the accuracy of ourpredictions is the force positive by the true
positive rates actually the area under thatcurve. And it's quite interesting because it
enables to measure, to compare experimentwith different units. It's just about how

(09:39):
much you're right or wrong about yourpredictions. So I'm not going to entern
into that table into details, butwe've looked into a competition. One of
them a sickly caa instance that wasacquired for forty million a few months ago,
and they're addiction ability in terms ofAEC for cytochrome. It's this enzyme

(10:03):
that metabolizes drug. Their areas isabout sixty seven to seventeen percent, where
we are in between ninety three toninety five percent. A random gas is
fifty percent. So then we've comparedourselves to online tools actually publications that were

(10:26):
that we're making better results and theonline tools and we consistently have done better,
and actually we realize that we startto be as good as the experiment
that we learned from. So oneof the first applications is virtual hyper cool
screen. So most formal companies theywill start a campaign to look for compounds

(10:48):
for specific target. They will startby screening let's say five or a dozen
compounds, and then will they willlacked subset of active compounds. So here
this is an virtual experiment where welearned on a data set that was looking

(11:11):
at into compounds like killing parasite Georgialambia and this is our testing set.
So actually if we looked at thetop ten percent of our predictions, they
still contain ninety nine percent of theactive compounds. So that means that people

(11:31):
could test much less compounds. Ifwe pre screen through our systems, we
have the potential to squeeze really thesize of a real hypoco screen to much
less compounds. So there is areduction in costs and in time. There
are this offensive example about a cytochromthis this enzyme like processing drugs inhibition,

(11:56):
and we had the active compound ofviagra in our testing set sinda field.
We correctly predicted that it was beingprocessed by that enzyme, but that was
not luck because our AEC is ninetyfive percent. So then we have been
also including like a property such ascelibility, lock peed, permeability. We

(12:20):
actually are able to predict most ofthe pharmacological properties that are being tested in
the pre clinical trials. So hereis again a virtual example about an experiment
that we would do to find compoundsagainst malaria. So we had a chlorocan,

(12:41):
which is a drug against man areathat was in our testing set.
We correctly predicted that it would beactive. But since our system is predicting
every property at the same time,we could see that at the same time
it doesn't affect any repair the nervalsystem or liver functions, so it could

(13:01):
be a good drug candidate. Thenwe also have been kind of rediscovering drugs
in the field of parasitology, bacteriology, oncology, and parology. So in
our testing set we correctly predicted thecompounds that are being used as drugs or
that are in tray as drugs.So the technology is therapeutic field agnostic.

(13:28):
You might think, okay, thisis very nice, but this is all
virtual. Maybe even is lyaying maybethere is a leak between the training and
the testing set. So we hadto do an experiment. It was during
COVID and since we predict many thingsat the same time, instead of looking

(13:50):
at only one type of experiments thatwas consisting of looking at compounds that are
inhibiting COVID in sarskof two in itself, I decided to look into also assays
that were reporting measurements of compounds beingtested for inhibiting host sell protein that are

(14:11):
being hacked by the virus or directlyviral proteins. We can also predict toxicity
through your essence, so that thatenabled us actually to successfully do our achieve
our first proof of concepts in thelaboratory. So I convinced the professor Alaly

(14:33):
in OLiS University to test six onlysix compounds that were actually our best bets.
Regarding all the compounds that were inin clinical trials or being used as
drugs, they were never tested forCOVID, and five out of the six
compounds were anti viral and uh noneof them were toxic at the active those

(14:58):
And now we have two of thesecompounds that are further in the pre clinical
phase. They work with different variantswith different types of cells, with long
cells, with patient cells. Sowe are in the process of patenting and
publishing. We've done some other proofof concepts that was exactly the same training,

(15:20):
but we gave a few other compoundsto Stanford University. They had another
redoubts that was about that that wasflower selves, but we were measuring the
same thing and we also discovered anew entire virrels. We have done that
other very different experiment. It wasabout compounds selectively killing cancer cells. So

(15:45):
here with the Victorian Center for FunctionalGenomics, we have had a little test.
So they gave us eighty percent ofa data set of about five compounds
that were either active or inactive interms of killing or not killing cancer cells.

(16:07):
And our predictions were testing on thetested on the twenty percent remaining compounds,
and actually we were as precise asthe experiment reproducibility. That's actually led
me to start some collaborations. Imean it's ongoing, but we are nothing

(16:30):
is produced yet, but I'm talkingto a few of the main guys in
sceneatics. Because if you are ableto predict out of a small data set
compounds actively killing cancerselves, why notsenacences. So this is something that's ongoing
and then we've just done something that'sthat's more actually in order to show that

(16:56):
we can offer this as a service, because we've been doing things on the
compounds that we chose, but actuallywe wanted to show that we can really
be useful in terms of lawge screening. So we commissioned a Syntems, a
Norwegian company, and the Norwegian Centerfor Medicule Medicine to test a library of

(17:21):
compounds so our system. Actually welooked into a library that was available for
them where they should be the largestamount of GD forty one HIV one inhibitors
SOP for twenty the protein of HIVone, and actually we believe there would
be about twenty two hits. Wecommitted to sending our predictions before they do

(17:48):
the tests so they can support usin terms of like the how good we
did, And actually we had twentyof the twenty two hits in our top
ten percent and two hits in inour top three predictions. So the probability
we get there by luck is aboutone or of a billion billions. So
the technology works that I've been talking. I'm starting to talking to pharma companies

(18:14):
and chemists were worried about how newthe molecules can be. So we started
to make measurements of distances between compounds, and we realized that the closest active
compounds in our training sets compared tothe predicted active is actually further than the

(18:37):
closest inactive, So that shows thatwe can really find compounds with new properties.
The Internet is not simply stupidly copyingsome motives. We are good at
generalizing and finding new families of molecules. So in terms of practicality, hydroput

(19:00):
screening is implementing. In the farmerindustry, you also have hyprod screening centers.
So people typically we'll test five hundredthousand compounds. It will take them
like anything between three weeks and sixmonths. The cost is about one million
dollars, and what we can dois to propose them to actually screen and

(19:22):
ten times as compounds, learn fromthat data, and then virtually screen not
only five dozen compounds, but allthe compounds that you are actually able to
purchase. So twenty three million isactually the reliable providers that we contacted that

(19:44):
we knew were reliable and we werefast enough to send us the data so
we can implement any new library.Also, we have this collaboration with one
of the main molecule providers and theyhave a data set of virtual a virtual
library of miracules that they can actuallybuild, and we've been creating a system

(20:11):
Denovo compound generating system that enables usto put the fragments that they have together
and to reach it's in the rangeof fifty billion compounds, so we are
sure that for customer research, thecompounds that we want that we virtually screen,
they can be then physically produced,so that's pretty important. Then,

(20:37):
just an idea about the scope ofthis, like the accuracy of our prediction
is really what we focused on,and as you can see, the prediction
accuracy is increasing the number of heats, so we can clearly show the value
of our service because the heat isin between fifty to fifty K to to

(21:00):
two hundred K. But with theaccuracy we have that's on par with experimental
accuracy. We have about five timesmore hits than what you would do traditionally,
and these numbers are based on thestatistics that were actually correct with our

(21:21):
proof of concept that I should before, so we're launching now service that's a
little bit easier because the virtual Hydropursescreening will actually force us to get data
from from the customer of our customerbecause hydrocul Screening Center is doing it for

(21:45):
a third party, and even thoughwe don't need to know the target activity,
it has been a little bit complexbecause it's a three party contract.
So another thing that we can dothat we're happy to do for any biottech,
especially working in the launde VT field, is that they can just send

(22:06):
us a file with the with thecompound structure. So there is a very
simple way of doing the name ofthe of the way to encode the compound
is called a smile and out ofthis we're able to predict most of the
pharmacological properties. So if they havea list of compounds, they can already
pre select the compounds that will notbe toxic and that will have the good

(22:33):
pharmacological property. So that's very interestingbecause that's just a part of the of
the service. But what we cando since we can screen ten thousand compounds
per second per GPU and in abillion per day, we can not only
screen the molecule for its ability toactivate the target. But we are also

(22:56):
able to look into after target effectsand to sell compounds that will have that
will be the range of the goodpharmacological properties. For instance, if you
want some drug to access the brain, you want a good formation of the
blood brain barrier and so so atthe moment, we are working on two

(23:17):
things in terms of research and development. So we are curating our database for
the first technology, and I startedto initiate partnerships with the key figures in
the in the longevity field that haveaccess uh to to laboratories or that have

(23:40):
laboratories. And so we have targetsthat we have curated and we're about to
send our first longevity compounds to betested in vitro and then individual in c
legains. So we've been looking attargets such as TOR and R two,
hta, hip hr it short proteinand and so on and these uh these

(24:03):
are not my favorite targets, butthey were in our database and we have
good predictions for them. So thisis something that we are currently doing.
We have also assays for autophagy thatwe can predict very well. And uh,
as I mentioned before, we havesome some project with sceneriatics and I'm
trying to also initiate a projects witha few people regarding hydropod screening. Regarding

(24:30):
DNA repair is my big problem withour database. It's most of the the
DNA repair targets people are looking forcompounds that actually screw DNA repair because they
want to kill cancer cells. Soso we will probably be able to find
some compounds with the negative results,but that's a that's a little bit tricky.

(24:53):
So I'd like to have a projectwhere we can generate UH lots of
data and let's see a high footfoot data level for to find new compounds
that we'll be able to to thatwe're enable to trigger a reapair. Also,
we are creating everything that's end toeduality these orders. So usually the

(25:15):
way we collaborate, we've done thatwith stand Forward and we always UH we
basically tell them here are compounds thatwe can send you for free, we
test them for free, then weshare the IP and out of this UH,
if something happens, we can startwriting publications, patents and then look

(25:37):
for money from grants or basically createthe spinoff. And most of the people
we've done the proof of concept.We are doing such things with us at
the moment. Then there is aonce we get some more funding, we
would be able to work on transcriptomicdata that would be very interesting and also

(25:57):
on a high content. But that'sanother story. This is really research and
development. There is another research anddevelopment I thing that we're doing which is
actually advancing very fast and and excitedabout it because not long ago we just
had a prototype that we call aconformer that folds small manicures faster and more

(26:22):
accurately than the state of the art. So our approach is to learn from
simulations and then to kind of usethat data to to feed our deep doer
nets in order to do good predictions. And actually here we optimize for the

(26:45):
energy above ground state. Uh,and you can see that we are better
than the state of the art.So we predict lower energy for the conformation.
And now we have a docking prototype. We're about a thousand times faster
than current tools, and our preliminaryresults show that we are as bad as

(27:10):
them. Actually, but very soonwe we should be able to have a
working tool because that was a nearprototype, and we have several things that
we can optimize on. We basicallyimplement some some some force fields that are
approximations of approximations of like shooting theshooting as equations, and we are we

(27:33):
are choosing better force fields and we'vedone that we kind of shallow nets compared
to our deepnal net. So verysoon we're very confident we will have a
docking system with an unprecedented accuracy.So now that's that's probably more of a
pitch side. But people would tellme, Okay, what's the difference between

(27:59):
what you do when the others aredoing. Is that I think I want
to insist on the fact that weare the first to be as accurate as
the experimental reproducibility. And that's veryimportant because like the hypercus screening market is
eighteen billion. But then like thisis not only hyperpulse screening, because we
can predict for pharmacological properties, wecan do we can paralyze things and really

(28:26):
prevent lots of back and forth ofcompounds that are toxics, so then they
have to create new compounds. Wecan generate analogs of the of the hits
and then test them virtually before theyhave to be produced. Meanwhile, the
competition is surprisingly looking into for instance, generative AI, and here it looks
like you want to to generate let'ssay, look at a like and a

(28:49):
drugs as a key and the targetas a luck, and they want to
generate new keys, but actually theyhave no good way of predicting how the
key will interact with the luck.So with all this drug target activity guidance,
we we don't think that it's reallyefficient. Also, there is lots

(29:12):
of language models. I mean they'repowerful like in other areas like judgeing BT
we have seen, but with languagematters you're unable to see the difference between
your left and your right hand,and so you can do nothing about key
reality. Then regarding docking, peopleare still looking into crystallo graphic data,

(29:34):
but crystallographic data is very scarce.You have maybe forty compounds for one target.
And then also the it's not soaccurate because the compound is not stable
that way, it's like moving constantly. So and finally, forty compounds per

(29:55):
targets doesn't help you. With deeplearning, you can and do nothing.
So here by generating data with withforce fields uh, and then then kind
of squeeze like learn from that andmake this computationally faster, we are able
to harness the power of deep learningto already make good predictions. So in

(30:22):
terms of a business model, Imean, I don't know if this is
really in the scope of like thisthis talk, but so we we are.
We are basically trying to work withpeople in the longevity field and co
develop a pipeline of longevity compounds.So so far we are we are starting
to send our compounds to be testeduh in different areas, So I'm shooting

(30:45):
people who work in different areas oflongevity, so like it would be like
scenerlytics and a repair any specific expertof some targets. UH. Then also
we need to to feed the companies, so we're starting a service where we
can virtually uh test compounds for theirpharmacological properties. We can also do the

(31:10):
trick with the virtual and hyprocus screeningby learning on much less data and or
around people, so to test tentimes less and get actually more hits.
And I also initiated some conversations withthe with the pharma industry where I would
like to have more of milestone basedproject with a smaller front payment. But

(31:32):
since since we know the technology works, we would like to have like some
kind of success fee when the compoundsare reaching a specific phases, so we
will focus mostly on the on theprachinical of course, because after it's a
it's another story with the with theclinical trials. But still I think the

(31:52):
time for the whole prachnical phase isabout two to five years. Here we
could squeeze that to five months actually, and regarding the what we have done
for COVID, usually it would takea few years for a few groups together

(32:15):
to work on that. Back inthe time, we were two guys and
we worked on it for a fewweeks, so the cost is quite negligiable.
Then at the moment we're looking forseven million for two years just to
cover our custom I think we shouldbe profitable next year already. Uh.

(32:37):
And so we planned on being likeamong the main players with a better technology
but still less resources. So ofcourse we need to spread the world and
put a foot in the farm industry, which is quite difficult and lots of
the budget. We're willing to actuallyhaving business development guys who have a food
in the farm industry to just openthe door to give us a chance to

(33:00):
show that our tech works. Solike the take on message regarding the technology,
So regarding what works and we canalready offer as a service or that
we're doing through collaboration, is thatwe have a technology that can literally save
hundreds of millions and years in thedevelopment for each drug. It is proven,

(33:20):
it has been tested in different laboratories. It is fast and accurate.
We can test ten thousand ricules persecond. It makes one billion per JP
per day. We learn at everyproperty at the same time, so we
are able to paralyze heat discovery.We lead optimization, so not only we
will look at a compound that's patent, but we will be able to select

(33:45):
by the probability of being toxic andnot having some specific off target effect and
so on. And then the technologyis in any I mean it is therapetic
fielm agnostic. But the idea,of course, since since my main objective
is to survive, all what wedo in the house is to develop our

(34:09):
longevity pipeline. So I think that'sit for the short presentation. So I
guess you guys will start asking mequestions. I'm sorry there was something in
the in the chat there. I'mnot sure if something was for me,
but I didn't read anything, soyes, so there are some interesting questions

(34:34):
from our viewers as well as ourpanelists. So I will pose this one
from prosper who wonders, can perhapssome of the compounds that you are testing
virtually lead to a cure for HIVas we know thus far, essentially the

(35:00):
options that are available for people aretreatments. They can lengthen the life expectancies
of people who are afflicted with HIV, but not quite cure the disease.
Do you think there is a compoundout there that just needs to be subjected

(35:20):
to testing, needs to be identifiedthat could actually provide a cure. I'm
not sure there is a compound thatalready exists, but definitely with our system
we should be able to generate acompound that we do the job. Because
now with our small tests with theGP forty one HIV one protein protein that

(35:47):
helps HIV want to enter cells,we have been we have been able to
find a patent nover anti virus.So I believe that it it would definitely
help and it would be possible.Of course, we never know because this
is R and D. But myfirst guest would say yes, yes,

(36:12):
thank you, and along these linesin terms of the process involved in using
this approach to synthesize new compounds ascompared to testing the effects of existing compounds.
How many more steps would be involvedand how much more time would it

(36:35):
take if you were looking to synthesizea novel compound and use this approach as
part of the initial identification of thestructure of that compound, depending on what
you wanted to do. Two weekswith the provider that we were talking to.
Nice, Nice, So there's notthat much of a difference between evaluating

(36:57):
what already exists and creating something completelynew. And it's also interesting to me
that you mentioned there are some naturalcompounds that you've been interested in studying.
Do you have any insight as towhich ones are actually effective or promising and

(37:19):
is it the majority of the naturalcompounds that are sometimes advertised or just a
few, just a small subset thatactually have beneficial effects. Okay, so
we've not done our first screening regardingregard our Lungevity pipeline for natural compounds.

(37:42):
It is in the process or withina couple of weeks, we should have
that. But I've made an effortbecause I want something to be quickly on
the market. So we have allthese libraries and we have kind of classify
them by natural compound, f theyapproved natural compounds, natural compounds, drugs

(38:07):
that are like in clinical trials orthat are approved, and any compound that
you can purchase physically from a library. And finally, any compounds that you
can synthesize. So all of thesewill lead a like very short term,
medium term and not too long termbecause we really hope to accelerate the whole

(38:30):
process. And so, yes,and there will be some compounds during our
HID thing. We've seen some somesome fancy compounds occurring there, So no,
there is uh, there will besome natural compounds. I believe they
are not going to save our lives, but they might be enabling us to

(38:53):
reach that escape velocity. I don'tI don't talk about this this way,
but it's actually have the same visionabout it. So the faster we can
eat something that can makes us leavesignificantly longer, the better it is.
So I will definitely look into directlytrying to work with the food supplement company

(39:17):
to directly implement the compounds in thein something that's commercializable. They should be
better than what we have today.Yes, yes, indeed, and anything
that gives us a few additional yearswould also increase our probability of reaching longevity

(39:39):
escape velocity. So it is allin the interests of moving in the right
direction. Now, Alan Crowley hasan interesting business related question. He notes
that your company is organized as aGmbH, which in German speaking country is

(40:00):
essentially quite similar to a limited liabilitycompany, and he wonders how does that
influence national and international investments? Notthat much actually, to be honest,
I did set up that company duringCOVID. We didn't know where to go.

(40:23):
My business partner was homeless in betweenSouth Korea and Germany and Hungary.
I was in Hungary. We sowe needed to find a way to set
up a company where we would have, let's say, the potential to attract
some talents and where something is happeningin terms of in terms of biotech and

(40:49):
pharma. So I think that asa German company in Munich, it's quite
a good image and let's investors thatinvest internationally don't have problems for that.
However, I know there are somespecific grants that are only for American companies.

(41:12):
I mean, initially I wanted todo that in the States, but
during COVID was impossible. So soyeah. For to make the answer short,
it's not really a problems as aGerman company. Specific investors would only
invest in specific areas, but mostinvestors in that area are probably really let's

(41:37):
say, area agnostic. Yes,thank you very much for that answer.
And we have a question from DanielTweed who wonders to what extent do you
let pathways to marketability drive your researchdecisions versus doing promising basic research. It's

(41:58):
a compromise. So I was notso interested in h in the COVID,
but it was easy to do theIt's just some some people at the beginning
with thinking, oh, yeah,but this technology is just for repurposing.
No, it was for repurposing becauseit was cheaper to purchase the compounds that
already exist than to synthesize new onesand uh and we had of course less

(42:22):
chances of them to be toxic becausethey're already on the market for other purposes.
So we will always have to doa compromise because if you want something
that that that actually leaves, itneeds to exist on the market. So
we were really striving to do somesome some fundamental research. But of course

(42:47):
we we need to we need tolive to survive also to make some noise
or it is a compromise. Yes, fair enough, and I think every
research oriented business makes that compromise toone extent or another. Daniel has another

(43:07):
question. He wonders when was quantitativestructure activity relationship modeling first developed and approximately
how large is that community. I'mnot a chemist, and I know that
has been there for a long time. But the technologies that are used in

(43:32):
terms of like understanding the data,it was not even machine learning or really
absolute machine learning to like shallow learntor like random forests, super vector machines,
and so it kind of kind ofgave a bad reputation to quisor and

(43:53):
and people. They did not giveup, but in the farmer industry and
also in in in other research areas, they just don't emphasize so much on
it and they use it as anot as as a guide, but just
like like just like something that mightgive them some insight. Meanwhile, if

(44:14):
you reach experiment that accuracy, thisis like the main tool that people should
use. So but that was reallyenabled by our hard focus on being very
precise. That so we had toto let's say, get inspired by by
vision recognition and to use the latestmachine learning model. So actually Q s

(44:42):
our itself. It's like like predictingan activity out of a molecule structure.
But the way you predict it isuh, you have a million ways of
doing it at a midion, manyways of doing it. And I think
we found the strategy to make aqsor useful, but I have no idea.

(45:05):
It must be something like twenty years. But it was basically useless back
in the time because the statistical methodwere not good enough and there was not
enough data. Now there is moredata and it will start to be I
mean now for us, I believeit should be the first thing that you

(45:25):
should do in terms of screening.Very interesting how the availability of data makes
techniques that were previously developed but perhapsnot had as much application quite a bit
more useful now, So thank youfor that answer. And Daniel has one

(45:47):
more question, what computer languages doyou mostly use? And as Python suitable
for those kind of work. Yeah, so this is actually a language that
many people know and have at leastsome basic experience with, and it can

(46:09):
be harnessed for such quite remarkably advancedapplications. So thank you for that answer.
Now. I was also curious toget your thoughts on Googles Well,
Alphabets alpha fold deep Mind is asubsidiary of alphabet that released this artificial intelligence

(46:34):
system called alpha fold for the identificationof protein folding configurations in a much more
effective way than the previous approaches,which were essentially using computational brute force and
distributed computing across millions of individual userscomputers and alpha fold. My understanding is

(46:59):
it's not perfect, but it canaccurately predict a lot of protein folding configurations.
So when you compare and contrast yourapproach with alpha fold, what can
you say about that? So Ilike the concept of alpha fold a lot
actually because we will be able touse their technology in order to to look

(47:22):
at when our docking system will befinalized. We will use their technology in
order to have the the the threeD structure of our target. So that's
that's that's that's extremely useful. Butwhat they do actually is machine learning on
on on data that comes from fromsequences evolution, so it's it's very different

(47:49):
and there is not much about howthings are interacting, So it's a it's
a different technology, but it's atechnology that extremely useful for research and for
us you or talking once it willbe developed in a few months. The
main message here is that really alphafool. They're really good at making these

(48:14):
these uh these protein conformation, butthey don't look into interaction between these interesting
but I'm glad to hear that thereis some complementarity between those technologies. You
also mentioned how your technology compares favorablyto generative AI and large language models,

(48:37):
and one problem that even lay personusers have observed with those approaches is sometimes
they can just make things up throughwhat's called hallucination, where they essentially,
uh create facts that are not actualfacts, but they seem like they might

(49:01):
be plausible, And to someone who'snot an expert in the field, that's
very difficult to distinguish those hallucinations fromactual facts. So that's a vulnerability that
I see of using generative AI fordrug discovery, because it could just invent
some sort of fictional compound that looksvery good on paper, except that doesn't

(49:27):
exist and cannot exist. So howwould your approach prevent that kind of outcome?
Okay, so, actually regarding ourapproach, we are creating compounds,
but at the moment, we reallywant to make sure that we can synthesize
them. So we virtually screen composite. There is some kind of evolutionary system

(49:52):
that will enable us to screen evenmore than one Videon compound per day.
Actually, we we build the compoundsin a way, in the same way
that they are built by the companiesthat they can that can produce them.
So actually we can come with likevery large number of compounds, but we

(50:14):
make sure that people can synthesize them, because otherwise it's it's will it will
take too much time. Yes,and that makes sense. What kinds of
criteria do you consider when you thinkabout whether a compound could be synthesized.
We don't really, I mean,there are there are two things. I

(50:39):
mean, there are systems that showbasically if if if the thing would explode
or not and uh and and onthe other hand, like we just stap
into libraries of compounds that either existor that's uh. The chemists who are
actually offering the service of synthesizing,like in its virtual database of lots of

(51:02):
compounds, we make sure that wesee that as in that can you catch
space? Yes, thank you.And more generally, when your system predicts
certain effects for a compound, whatkinds of empirical tests are necessary to ultimately

(51:25):
determine the validity of that prediction.So for a lot of the properties,
you said, your system has accuracyrates in the eighties or nineties of percentages
of what actual tests in a labwould ultimately confirm or disconfirm the prediction.

(51:50):
It's it's rather in between nineteen andninety five, and it's the AAC.
And then actually we're we're getting betterwhen we send compounds because we as as
we have done with COVID, weare actually selecting upon different predictions. So

(52:10):
we have we have built that thatthat system that actually allows us to make
an algorithm of all the predictions.So we don't only look at experiments that
look at is this compound active forthis, but also other experiments that that
can lead us to think that thecompound would be active as well. And

(52:31):
it's it's it's pretty simple. Itdepends on the it depends on the type
of experiments. So but it's it'sjust a question of sending the compounds into
into essays or into into cells andsee if they're active or not. So
you have you have like, uh, let's say, how can I how

(52:57):
can I explain this. It's it. It's basically standard assays, like it's
it's a it's a it's depending onthe experiment, it's going to be a
thousand euro to ten thousand euros.So it's it's nothing, it's nothing crazy.
You you just test and retest thecompounds and you make sure that you
have some some specific compound here thatis like usually active and some specific compounds

(53:22):
that are usually inactive. So youjust make sure that there is no problem
with the experiment. But it's it'sstandard as a measurement and it's not that
expensive, but it really depends onthe type of of assays that you're measuring.
Is for GP forty one, we'vedone that on a larger scale because
it was a molecular asset and itwas optimized for like small hypercude screening.

(53:46):
But for COVID, it was justcells that were like infected by SARZCO two
and the compounds are just grown themand you and then you look at at
the amount of RNA. But that'sit's pretty straightforward in simple. Yes,
thank you for that answer. AndI'm also curious in terms of what have

(54:12):
you observed the common sources of errorto be. In other words, how
can reality differ from the prediction,and why do these differences exist even though
you're trying to minimize them, andhow can they be minimized further? So

(54:32):
far, I mean, we we'rekind of happy to be as good that
the experiment that accuracy, and thenwe did not see any patterns in the
compounds that were let's say, inthe few compounds that were not considered as

(54:53):
active, and I think it's isuh that that was falsely considered active.
Unfortunately, that will most of thetime, I guess the experimentation related.
So if our experiment is very accurate, we will be very accurate, and

(55:13):
we might even reproduce the same typeof error as the experiment, But we
have not seen that so here,I'm just like thinking in real timement.
So far, we don't have anyany direction into specific types of error.
We're just happy to see that ourresults are reaching experimental reproducibility. If we

(55:35):
have in our data, and if, of course the training DETA is not
an analog of the same compounds,we need compounds that are a bit different
from each other, so the AARNetcan build a law between the compound structure
and the activity. Yes, thankyou, and you spoke about also evaluating
the off target effects of a compound. Could this be used to, for

(56:01):
instance, discover hidden side effects ofcertain drugs, even drugs, for instance,
that have been on the market fora long time. But sometimes we've
seen the FDA recall certain drugs becauseside effects were discovered several years after they
were approved, and those side effectsended up being a lot more severe than

(56:25):
previously thought. So could this approachperhaps shed light as to perhaps hidden adverse
effects of certain drugs that were notconsidered before. Yes, the objective is
to do that in events, ofcourse, because you don't want people to
have the side effects. But definitelywe this is something that we could do.

(56:51):
Yet we are actually creating our wholedatabase, and then once we have
all the targets that are usually eatingscreen, we will be able to predict
any standard of target effect. Wehave already some of them. You know
at met service you don't want likethis. There is a specific recept or

(57:15):
I g ORG that's going to createlike some Hart disordered reaction, and that's
something that we can basically prevent,predict and prevent. Yes, and I
think that would be of great benefitto a lot of people. Our friend

(57:37):
Deger Cornell has a provocative question.First, he says, your presentation is
interesting and clear. He wonders whynot work in collaboration with in Silico Medicine,
which is also using artificial intelligence fordrug discovery, rather than be a

(57:57):
competitor. He thinks, it seemsyou have the same goals and motivations,
or I wonder if you have adifferent perspective, perhaps your areas of focus
are different in some way, oryour systems are different in some way.
But what do you think of thisquestion? I have not. I have

(58:19):
not. I have nothing against inCilico Medicine. I just don't believe they
have the right approach and they're spendinglots of money. I mean, this
is not it is not impossible.I mean I wouldn't completely refuse a collaboration

(58:42):
with people in the area since theobjective, the the ultimate objective is to
accelerate drop discovery for longevity. Butso far, I mean it might be
I mean, I'm not saying no, I'm saying it's a good idea.
It might be dangerous from a businessperspective, but yeah, it's nuts.

(59:06):
Yeah, all right, Well,thank you for I know what. Yes,
yeah, he has been a frequentguest on our Virtual Enlightenment salons as
well, and he always asks interestingquestions definitely worth considering. So now I

(59:29):
will go to our panelists and seewho has additional questions for if i'm let's
go to Ben Bolwick. I wantedto follow up about alpha fold. I
think I was speculating that you wouldn'tbe able to practice or to even have
cortex without alpha fold having come along, but your answer made it seem like,

(59:52):
ah, you'd have been just fineif alpha fold had hadn't existed.
Could you clarify how useful alpha foldis for you or if it's just like
our system doesn't use that at all. Okay, so the first technology does
not use alpha fold at all.Like we traind from data, like the
input is americule, the output isa property and uh, and we're good

(01:00:15):
with this. But for targets forwhich there is no data available, you
need to either create the data throughhigh input screening or proceed to docking.
Ducking doesn't work so well today.So so for proteins where you know the
structure, we could actually use crystallographicdata, but there is no pressive crystall

(01:00:38):
graphic data for all proteins. Soactually alpha fold will be useful for us
to expand the docking to any anytarget that actually is interesting for us.
So it would just like give usa big hand for for the the extension

(01:00:59):
to uh uh to the to theducking to uh let's say born on targets.
Yes, thank you for that response. And now let's see if art
Ramon has a question. Yeah,yes, And in this creation of molecules,

(01:01:24):
do you ever run into a moleculethat has a really negative effect it
just rips apart sales or anything likethat. That's novel anything that comes out
from from your researcher. Uh,we have not really. We were rather

(01:01:45):
digging into selecting molecules that are notvery toxic, uh, but definitely uh
regarding uh I mean no, no, we we we've not encountered this,
but this is something that could bethat could be found. Yeah, yeah,
I mean the idea is that ifyou want to selectively kill cancer cell,

(01:02:07):
you know, on something too aggressiveto kill all the cells that are
instagraunding the cancer. So yes,thank you, Daniel. Yeah, really
great information. How compatible do youfeel your work might be with upcoming advances

(01:02:29):
in nanotechnology. Molecular docking sounds anawful lot like molecular level machining, And
is that something that you feel yourposition to take advantage of if and when
that becomes more feasible. Yes,this is something I clearly have in mind
from the beginning that we started towork on ducking because these molecular machines.

(01:02:53):
We are molecular machines, right,So because people have this idea of like
small robots that are metal or whatever, but I believe that the most efficient
nano robots will be kind of builtthe same way that we are a build.
So actually, like predicting how goodsome specific shape is going to intercalate

(01:03:14):
into two specific bases, or likeamino acids or enter into some specific receptor
is going to be useful. Sodefinitely I plan on being involved in in
such projects when the opportunity is here, when the talking system is working.
Do you think that kind of nanomachinerepair is the holy grail of longevity perhaps,

(01:03:37):
or any opinions on when we'll getthat date wise, well, so
far I would. I would lookfor compounds that help to increase the repair
that already exists, then new repairmachines. I mean, it's not science
fiction, but it's it's uh,it's not in the in my plan for
the next let's say, five years, but but it's it's not nice for

(01:04:00):
sure. Yes that is right now. Yes that is quite intriguing, and
of course a step wise approach isreasonable. But I'm glad that you do
see that future of essentially nanoscale repairas being attainable, because that could undo

(01:04:25):
a lot of the damage at thecellular level in a way that can just
recur periodically, so people wouldn't haveto have like massive surgeries or other kinds
of let's say, potentially traumatic interventions. They could just have the nanobots,

(01:04:45):
biological nanobots repair their tissues on anongoing basis. So that is quite intriguing,
but perhaps not in the next decadeor so. D K. I
wouldn't say, I said five years, but like because things are happening,

(01:05:05):
but we we we don't know.But so far, I guess that if
we find better synolytics and we canlike harness some kind of like healthy young
cells, like like destroying the world, bringing some some young and and then
like also giving medication to the bodyto to slow down the aging through various

(01:05:28):
like strategies such as like reducing activitiesstress increasing the general repairents on is the
is the first step, and thennanotech is the step after docking would help,
like especially like nanovods that would bemade out of like let's say biological
let's say organic molecules. But Ithink the definition of nanobots, uh,

(01:05:57):
not nanobots, but like I thinkwe will go really close to the medicular
machine that we have and just enhancethem a little bit. So we will
it be a nanobot or not,Like that's the question of definition. But
you may imagine that you do acrunch gene with the DNA ripper system that's
a little bit enhanced. Is thatthe nanobt So just a question of definition

(01:06:23):
because there were not too far interestingSo we may have a kind of subtle
emergence of nanobots when molecular machines essentiallyprogress to that point where the distinction becomes
blurred. And Alan Crowley writes inour chat, yes, doctor de Weber,

(01:06:48):
science in the world is better thanscience fiction, so thank you for
that comment. Alan and Deguer alsosays thank you for the answer. He
writes in general, I think morecooperation, open source activities, and less
competition are better for longevity. Thisis also coming from his left wing idealist
side. I think it's worthy tohear some of your commentary on this.

(01:07:13):
I actually raised this question at thefirst Longevity Summit Dublin in twenty twenty two,
where I observe that up until nowthe longevity field has been rather collaborative
most often. There hasn't been let'ssay, cutthroat behavior observed in it,

(01:07:35):
quite in contrast to what has happenedin some related fields. For instance,
in the field of mRNA vaccines,there is now quite a strong patent battle
between Maderna and Pfeiser, and Madernasued Peiser in the middle of the pandemic,

(01:07:57):
essentially alleging patent infringement. So clearlythere wasn't an attitude of having a
shared goal and trying to solve thesame problem using a collaborative approach. There
was more of an attitude of essentiallygetting ahead of one's competitor at all costs,
even if it risks undermining the progressof the field. So how do

(01:08:24):
you think it might be possible tomaintain a more collaborative outlook within the longevity
community, so that we don't havesome people essentially undermining the research of others
or making it more difficult to makeadvances and commercialize advances. Yeah, undermine

(01:08:46):
the projects of others is like nonsenseif you really are in the field of
longevity and you really believe that,I mean, if you want some results,
because I mean, I'm happy tohelp any company developing some drug if
we you know, like with anytype of like the prediction that we can
predict and so on. But Ihad this same type of questions from DDA,

(01:09:12):
but I also wanted to ask forredaly something just that it's not really
linked to lungevit. But in general, if you look at the landscape of
DNA sequencing back in the time,it took like fifty years for the world
and like over a billion to getto the sequence, and Craig Venter got
about two hundred million. And justbecause I thinks were well organized in let's

(01:09:38):
say, much smaller but more organizedstructure, he was able to do much
faster for much cheaper. So actually, also if you get the money in
the right hands and it's done theright way, it might also be better.
So the idea is that at somepoint, maybe some companies have betweening
us if we believe that we canbe better if we get some part of

(01:10:02):
the of the stake, then todevelop faster to something new, it would
be it would be beneficial to keepsome of the data not open source.
However, I'm very happy to collaboratewith with with different companies and uh and
to share data with them. Andwe are doing that like training from the

(01:10:24):
data or like sending our compounds tothem. And uh, I think in
terms of drug discovery, UH,especially from because we don't have the vision
of the farm out the chemist peopleyou know, a geneticism may business potary
quantum physicists and uh, actually wewe uh we believe that the scope that

(01:10:46):
people are happy about their libraries ofcompounds, but so many compounds that you
can make. It's a little bitcrazy to be attached on some compound or
compound family because we can do likealmost an in need number of compounds.
So I don't see the competition beingthat hard in that field, especially that
people who are fighting towards the samedirection. So I would I encourage more

(01:11:11):
and more collaboration. But I'm juststill saying it's like some data, of
course, for the survival of thecompany cannot be open. So yes,
and that is fair that a certainbalance needs to be determined based on the

(01:11:33):
revenue model of the company in question. Alan responded to DDA and said that
he is with DDA to a degree, but we do have to make longevity
pace so that people who will pursueit. So a similar thought, Alan
also recommends a book called Where GoodIdeas Come From. He says, this

(01:11:55):
book does an excellent job of showingwhere collaborative versus individual and secret versus open
research produced the best results. Andhere is a link to this book.
It can be found on Amazon.The author is Stephen Johnson. So I

(01:12:16):
would encourage anybody who is interested inevaluating the relative merits of these approaches to
check out this book. Alan,would you like to elaborate or perhaps ask
a further question of Yvonne Well?Sure, I'd like to elaborate on the

(01:12:39):
book Where Good Ideas Come From?Because I thoroughly enjoyed it and I recommended
I recommended it multiple times in thisform, and it basically has six chapters
about influences and aspects of existence thatbring about good ideas. I mean it
uses a continuous metaphor throughout all thosesix chapters in the and finale that it

(01:13:00):
has, which is uh goodness,the guy that invented UH evolution. Name
is Casey right now. But becauseJohnson did a study of his journals prior

(01:13:20):
to developing the theory of evolution anduses that analogy along the way to show
how sometime to take several different factorswhat he calls the slow hunch and not
the aha moment. So you know, this building upon of ideas over time.

(01:13:41):
We like to characterize it as theyou know, the Aha moment when
we say, oh, here whatwe need nanobots, whereas actually we started
with industrial you know, industrial giantindustrial machines and slowly progressed towards anyway.
The entire book is fantastic. Inthe last concluding chapter does show a relationship
over time and a matrix that involvesthose variables about when and where those things

(01:14:08):
are work best for innovation and developmentof good ideas. That's the book.
But going on, I really I'vegot a I think it's a fan.
I've got a question for you.I think your work sounds fantastic, Doctor
Baber. The this idea that you'reyou're researching the proteins. It sounds to

(01:14:29):
me almost like, uh that isthe first step before finding out combined results,
the pharmacological results. So what doyou have a like, what's the
pipeline? Do you like, providethe seed of an idea, here's a
there's a protein that can be developedand studied and turned that over to another
organization, or do you care asyour organization follow that all the way downstream

(01:14:51):
to the speak Okay, So unfortunately, uh no, First, we are
only at the moment working with smallmolecules, which is most of the drugs
and so small macaules that will activatetargets that could be proteins or that could
be a bad way because sometimes wecan do predictions without even knowing the targets.

(01:15:14):
So U to to let's say targetan effect and actually that's the idea.
I mean there is something that wewill let's say that have open source
like our code for AI, becauseI mean we can we will not be
able to eat without it. Buton the other hand, I see,
I know that I won't be ableto be bigger than ten times fizer within

(01:15:40):
the next decade in order to producedrugs to make myself live longer. So
I gave up that idea since theyears. So the idea is really to
have the compounds that have a veryhigh probability to work, and to send
them to the right partner who isable to validate that these come on work,
and then with these partners look forother like let's say mostly financial partners

(01:16:06):
to push that further in order tothen get some other funding to go through
the clinical So really my objective withinthe next next the same four or five
years is to have say two hundredcompounds in fifty companies that are actually progressing
like at the end of the preclinica or like already advancing in the clinical

(01:16:30):
trials, which with a much higherprobability of success. So that will not
be done by one entity. SoI completely agree here, especially for LONGVIT,
we need the collaborative approach. That'sthat's it's just necessary. So and
then that's our strategy, thank you. Yes, indeed, and that truly

(01:16:51):
does speak to the value of acollaborative approach because when the entire longevity field
makes advances and prospers, we allwould benefit. Deguier writes in the chat
He's sure that private organizations can bebetter than public organizations. However, if

(01:17:15):
you do not share the results,it is not useful. And theory patents
are for everybody and only profit isprivate. Practically, however, he writes,
patents are often made to not shareknowledge, and he says to Alan
that actually longevity would pay by givinglongevity to people and extraordinary fame for those

(01:17:40):
who find something useful. Well,I can say to Deyer my own motive
for being in this field is clearlynot monetary. It is to not die,
because I see life as the greatestwealth, and of course the longer
I am alive, the more opportunitiesI have to do anything whatsoever with my

(01:18:01):
time, including making money. Now, I am not in a position where
I am running a business myself,though, but sometimes, as you pointed
out, Yvon, businesses also benefitfrom collaborative approaches. And to the extent

(01:18:21):
that patents exist, I agree withDega they should be encouraging innovation rather than
discouraging innovation. So a question arisesfrom this, have you seen a situation
in the longevity field or in arelated field where somebody was holding onto a

(01:18:44):
patent and preventing others from utilizing it, preventing others, even if they wanted
to say pay a small or reasonableroyalty, from actually using the technology or
the method that that pattern covered tomake further progress. I've not seen that

(01:19:08):
myself. There was just this guyat some point who who kind of like
stalls some institutes and companies money becausehe bought some patent and beers. He
won two genes for sequencing, butI think was extremely buried and it worked
too long, and I think itwas a scandal because this is just for

(01:19:30):
diagnosing. I mean also like it'sreally predictive in terms of of of of
breast cancer tests or so, No, this is not something that that I've
seen besides that specific example. Soyes, it's going to become more farma
because before like longevity was snake oil. Now it's getting real and so we

(01:19:53):
will be like fighting the same fightas pharmas. So there will be some
some struggles. But the idea isthat at some point, if you have
some ideas, some vision, someplan, you need some cash in order
to push that plan to the tothe reality. And to make that cash
need to make some money, andif you don't have any ip, you

(01:20:16):
won't be able to get that money. So but of course, like my
strategy is mostly collaborative, and Iwould be happy to have a competitor finding
longevity compounds. Thanks for collaboration withus, even if we're just a service
provider and get almost nothing out ofit, because I'm going to eat the

(01:20:39):
compounds. So like that's the thebig difference. I'm just saying I need
to keep a little bit of theIP in order to grow. But then,
like, of course, the wholework, the whole development is collaborative,
Yes, indeed, and that's understandable. Alan writes that, yes,
the benefits of health span and lifespanare valuable rewards, but the research is

(01:21:00):
expensive, and Ellen says he wouldgladly pay for better memory. Mike Lasine
writes that automakers and others have donethis kind of thing for years. They
snatched up patents and inventions and thensat on them. So this is more
of a problem in more established industriesor more conventional industries. Big pharma has

(01:21:26):
a bit of that problem as well. But I'm glad, Yvon that you
haven't seen that as much in thelongevity field, which speaks to some of
the unique dynamics and incentives in thatfield. I just hope we can keep
it more collaborative and we don't geta situation where all of this research essentially

(01:21:51):
falls within the power of some conventionalpharma executive who then says, oh,
we can just sit on these patentsinstead of allowing this set of discoveries to
come to any sort of practical application. There are some who say that big

(01:22:11):
pharma is incentivized to maintain the statusquo and sell people essentially maintenance medications for
their conditions, but they are lessinterested in pursuing cures. Or if they
have a blockbuster drug that kind ofmaintains people in a tolerable state of health
but doesn't cure the underlying disease,and there's this new compound that could cure

(01:22:35):
the disease for good. Some peoplesay, well, the pharmaceutical company's incentives
would be to get the patent forthat new compound and just sit on it,
not manufacture it. But what doyou think about that? Do you
agree or disagree that that would bea powerful obstacle, or do you think
there's enough of a profit opportunity thatbig pharma companies would more likely on true

(01:23:01):
longevity medicines or true disease cures andprefer them over the maintenance types of medications.
I think if they have something thatworks better and that they can finance
it, they will go for thebetter products. But definitely they have to
sell what they produced, so andthat's why sometimes you are a little bit

(01:23:25):
surprised about the little amount of sideeffects that were kind of discovered during the
trials and then what's happening in reality. But definitely I believe that if they
had such a compound, they wouldlike try to push it further because there
is basically more money in a compoundthat will work better, and it's also

(01:23:48):
it's also marketing and reputation, SoI don't think they would hold on to
compound too much. And actually,as I'm saying, compounds for me,
actually don't that very important. Ithink there are so many compounds. What
comes is the technology that finds thetype of compounds at work. But one

(01:24:09):
compound is not magic. If thereis one magic compound, you can find
another thousand magic compounds. These patternsare just basically protecting the guys who worked
to produce them, so they canhave a little bit of cashion continue doing
their research. That's the way Isee it, That's the way we were
doing. Yes, thank you forthat perspective. And our friend John h

(01:24:35):
writes that within his own lifespan,science and science fiction have gone from the
realm of the nerd to the generalpublic, and this is quite amazing.
The downside is that he, asan old sci fi nerd, feels so
much less special. And I wascurious, Yvonne to get your perspective on

(01:24:57):
this, because, like me,you were interest in defeating death from a
very young age. And compared toyour expectations of the progress that science would
have made by now, how hasthe reality matched the expectations or diverged from

(01:25:17):
the expectations. Are we further alongnow than you expected when you were a
child, or are we perhaps behind? Have certain advances not happened as quickly
as you would have hoped. Iwouldn't compare. I wouldn't think about myself
as a child, because I reallydidn't have much much data about I mean

(01:25:41):
to make myself a critical opinion,but I as a young adult, I've
seen things moving faster than expecting,so I'm more confident that it would happen
than I was before. I rememberwhen I started to university, I thought,
we'll find a way to get frozenand another second chance. And then

(01:26:06):
during the course of the university Idecided that it would be a way to
kind of avoid this like a coldsleep, by just like a belonging lifespan
until we find a way to repairus as as we decay, and that
is not too far, I believe. So if we can have the same

(01:26:27):
type of conversation in the same typeof physical shape in about twenty five years,
I think we're done. Yes,indeed, well I would agree if
in twenty five years we look theway we do now, clearly something will

(01:26:47):
have happened to enable that some sortof progress in medical science, and based
on the logic of the concept oflongevity escape velocity, that is pretty much
what we would need to do,and of course avoid accidents and self destructive

(01:27:09):
behaviors in order to not have anupper bound to our lifespans in order to
potentially live indefinitely. So I'm gladthat you are more hopeful now than you
used to be. You said previouslyyou had been focused more on the prospect

(01:27:31):
of cryo preservation, but now youseem to think that people of your age
would be able to avoid being cryopreserved altogether, at least in terms of
the probabilities favoring the outcome of justrejuvenating oneself. Would that be correct to
say yes, I mean, I'mgoing to fight for that. I'm going

(01:27:57):
to fight that to be in thecredit think as the idea. But another
thing regarding this is a little bitof a philosophical debate I had with my
business partner over the last thirty fiveyears. I've been a knew him for
that amount of time, and hewas really thinking that we would need to
understand biology with AI, and actuallyI was thinking that would be not impossible,

(01:28:25):
but very difficult. And I thoughtthat by finding some critical target and
get one medication out of luck onone of the targets, like Pharma is
doing these days, we would beable to get an extra fifteen percent that
will give us the time to findanother one that gives us another twenty and
so on. But I was reallybetting on luck, on that luck.

(01:28:48):
But this kind of like but here, like really I didn't think that the
machine learning would actually be as recessas we have made it with them being
as good as the experiments. AndI also do not believe that we would
use anything pertaining to the quantum Well, and I see that with our docking

(01:29:11):
technology, deep learning and fast computingwill enable us to actually harness that information
to be more precise in and inmodulating how a compound interacts. So these
are two things that are completely Imean I had them in mind, but
I gave them a very low probabilityof happening in that quest towards longevity.

(01:29:34):
But actually we are already harnessing theirpower. So it's pretty interesting. Yes,
indeed, And do you think thatquantum computers will actually become usable for
these kinds of applications, for machinelearning, applications for drug discovery? Right

(01:29:58):
now, there have been some successas quantum computing, but at a much
more rudimentary level, like making afew successful calculations. Do you think we
could have a quantum computer, saywithin the next decade, that would actually
be able to do a lot ofthe extensive work involved in drug discovery.

(01:30:21):
So actually, I mean here mybusiness partner should be the one talking about
this because he is the he washe's the professor in quantum information. But
what we have discussed is that ourdocking technology will be one of the first
applications of the quantum computers when theywill be there and really usable because the

(01:30:48):
way the information is coming in theway we will feed the Arnett is really
fitting. So it would be oneof the first applications. So it would
be very It would probab leaders tomake like unthinkable scale of simulations like protein,
like any type of you know,molecule interacting at a much larger scale.

(01:31:12):
So, and that will probably happenin the next decade. But there
I'm not I didn't follow the latestadvances and we have the rest of promises
with the quantum competing. It's goingto happen, but to the extent we
can use it right now, Idon't know. Anything between five to fifteen
years probably quite interesting. And ourfriend John h writes that quantum computing and

(01:31:35):
control nuclear fusion would be profound worldchangers, so technologies we hopefully all have
to look forward to. And yes, within five to fifteen years, that
seems like a reasonable timeframe for significantapplications of quantum computing to be realize.

(01:32:00):
Now, since you mentioned your businesspartner, there has been a bit of
discussion in the chat about his originalreply to you that humans would just be
replaced by robots. What is hiscurrent outlook on this? Has it evolved
since that time, especially since you'vebeen collaborating together on drug discovery. Yeah,

(01:32:24):
so, even though he's not themost empathic being as as as a
programmer who's really into his his thing, he kind of I think his objective
is also achieve like the goal ofreaching this escape the city status. He's

(01:32:50):
still a little bit not as confidentas I am, but I'm working on
a biology side a little more.But yes, that's the reason why he's
doing it, he told me.I mean, I'm not sure it's going
to work. Probably they are quitelow because he's a very skeptical guy.

(01:33:12):
But so basically he's not developing askindest he's working with me to do to
find medication for ngelity. Yes,indeed, And really what matters is that
people with his kinds of talents dedicatethose talents to making progress. And the

(01:33:36):
progress could be incremental. They couldhave different outlooks as to where it can
ultimately take us, but as longas they're working to take us in the
right direction, that's all good.So I'm glad that you were able to
essentially convince him to devote his abilityto this fascinating effort. Alan also has

(01:34:05):
been very generous with his support forour virtual Enlightenment salon. He says,
thank you, doctor Weber, keepworking on the science that will benefit us
all you are making the seeds andpassing them on to growers. Here's to
you getting some of those fruits later. So thank you for those great words,

(01:34:27):
Alan. And now I wanted togive an opportunity for each of our
panelists to make either any final remarksthat they want to make or ask any
further questions. And we have theability in the next nineteen minutes or so
to still delve into some serious discussionof any areas that people want to bring

(01:34:53):
up. So who would like tostart, Ben, Neither of these questions
are awesome, So I'll just askthem both and you can choose if you
want to ask answer either of them. First, I was curious, besides
you and in Silico, if therewas another very interesting player in the machine

(01:35:13):
learning space for longevity medicine that wasthat you could tell us more about,
just for us to have a betterunderstanding of how the longevity field is using
machine learning in general. Across thefield then the other one. And I
don't know how relevant this is,but I was wondering about the regulatory hurdles
that you face when you are testingor excuse me, when you are using

(01:35:39):
machine learning to get the data,when you're not actually generating data that's collected
on animals and then humans. Butmaybe you're so early in the process that
it's not really a hurdle that you'reeven concerned about. That's more like big
pharma and those that are getting itacross the finish line that they have to
concern themselves with that. So eitherof those or none of those or whatever.

(01:36:00):
Okay, Well, regarding competitors likethey don't really publish what they're doing
so much. We've seen that Recursionis acquiring some guys who look like they
are good scientists, and they areacquired sickly cap but they are they're looking

(01:36:25):
that they're doing they might do somethinginteresting. There is also atom Wise.
There are other companies that are Accentiaas well. They have received lots of
many but I've not seen any breakthroughtechnology so far in any of these competitors.
The most interesting thing that we haveseen is Alpha fold, but it's

(01:36:49):
not directly applicable to drug discovery,but we will honess it's its power.
So that's one thing. Regarding regulatoryissues, we're pretty happy about it because
we're just a data company, sowe get numbers and medicule structures and we

(01:37:09):
make predictions. We sell predictions,so there is no regulatory issues. We
really want to focus on the scienceand not on the paperwork. Yes,
thank you for those answers, AndI'm glad that at the level at which
you're operating there are not really significantregulatory obstacles because indeed, you are not

(01:37:36):
making drugs at this stage to provideto the public. You are essentially gathering
knowledge about what the potential effects ofvarious compounds could be, and that later
on could be used, for instance, to decide which drugs to do clinical

(01:37:59):
try trials on. But it's thatclinical trial process that's ultimately regulated by agencies
like the FDA. So you statedthat essentially it takes ten to seventeen years
to get a typical drug to marketthese days, and with your approach,

(01:38:23):
by how long do you think thattimeframe could be shortened? So I mean
I put ten to seventeen years.There are drugs that go a little bit
faster. So at the idea isthat, as I said, we were
not focused on your drug discovery,so like next year, we should have

(01:38:43):
new natural compounds that are already authorizedthat should be quite active, and then
republishing is taking three to five years, so these things, so we would
we would have like very quickly somenatural outcome. Was also in the frame
of maybe two years that they wouldnot need a d things, but the

(01:39:05):
ten to seventeen years. What Iknow is that from the two to five
years, we could really squeeze thatto let's say three months. That's the
pre clinical. After the clinical mightgo a little bit faster because of less
back and forth side effect cancelations becausewe're really looking into many of the of

(01:39:27):
the off target effects, and alsothe profile for cyto chrom ahivisions. So
people who usually take many drugs,they will not like make one citle chrome
tired at the same time as Iused two drugs that use the same enzyme.
So these are things that will slightlyaccelerate the process. Also, COVID

(01:39:50):
was pretty bad for many reasons,but it's kind of been able to accelerate
the clinical side because some of theholders were like a bush and they felled
forever, some of them just forCOVID, some of them maybe for the

(01:40:11):
best, but still it would acceleratethis. So I'd hope to have like
something on the market as a medigationanywhere within seems to be optimistic to eight
years, but before we will havesome rop up with compounds and natural compounds
that would be very efficient. Yes, thank you. So quite a few

(01:40:31):
years can be saved using this approach, which is quite encouraging. Daniel.
Let's hear your question or comment.Yeah. I used to do a bit
of public affairs radio college radio inthe nineties and one of my favorite guests
was doctor Bob Beck was doctor ofphysics in the Strobe Light and he had

(01:40:56):
a suite of electromedical and metallic ozonationtherapies that he claimed were capable of curing.
You can't use the word cure,of course, because the AMA has
a patent on that, but itwas basically colloidal silver, which elemental silver
has a natural antibiotic, anti fungal, anti viral property. Ozonated water,

(01:41:21):
which if you bubble ozone through icewater and then drink the water, you
can see your pulse oxygenation rates goup. From like you know, a
to over one hundred. The otherwas electromagnetic pulsing through a kind of a
coil which would you'd put over yourlymph nodes. And then the last thing
was a blood micro voltage pulsing biphasicDC And apparently it has been researched at

(01:41:50):
Albert Einstein Medical College and they weredoing not the in vivo version of this
where you put the little cotton wrappedelectrodes on your wrist, your break you'll
artery and vein, but an externalthing where they would actually take the blood
out and micropulse it with electricity.So are these do it yourself kinds of

(01:42:11):
medical technologies actively suppressed by the medicalcartel? You know, why does silver
have this antibiotic property? Why dothese micropulsations change the cellular potential for viruses
and pathogens to invade cells? Doyou know any work that's been done along

(01:42:32):
these lines or what's the promise ofresearching that? And can people do this
stuff on themselves, experiment on themselvesand be exempt from medical regulations? So
anyone can look up the Bob Becktherapy online. But it's again, it
hasn't reached a huge level of acceptanceor but apparently it's non toxic and non

(01:42:58):
invasive, so you know, thehypocritical says, do no greater harm.
You know, try the least invasivething that might work to cure a situation,
and then move on to something moreinvasive if that doesn't work. So
any thoughts on any of that.I'm myself taking about one hundred pills a

(01:43:18):
day and add some injuries, andhave done multiple treatment that were semi experimental.
So when things make sense and theyare not harmful, I wouldn't say
that you shouldn't try them. Thezinc thing doesn't sound too bad. The
ozone apparently either electro stimulation as soonto be beneficial in different areas. So

(01:43:45):
I've not investigated it personally, ratherat least not more than like for some
massive injury or like massive stimulation.But there think I used it for for
this infection. It's in many ointments. Silver was the element not saying as

(01:44:06):
well, Yeah, a lot ofelements, sorry, because I have I
have like some with civil and things, so that's why. And uh no,
no, no that there are easythings that work. I mean,
uh, like the right time insea is a good adjuvent for for chemotherapy,
and it should be used more forinstance, and and many clinics are

(01:44:29):
starting to do like injection of right. I mean, there are basic things
that can that can help, andif they are not harmful, why not
trying them? But the ideas there'sno money in them, so that's why
they're not being advocated or adopted widely. The silver is quite it's quite advocated

(01:44:56):
adopted, I'd say, like theright time in c has well. But
definitely uh, the fact the factthat there is no patentable thing there,
like is the treatment is not it'snot Uh yeah, I see, I
see where where you go, uh, where where you're coming from. But

(01:45:20):
I think more and more now withwith the social media, like natural medicine
and and uh the advantage of science, more and more people are experimenting on
kind of healthy approaches. So Imean, I if it's not harmful,
why not. But then but thendefinitely these uh, these things won't be

(01:45:44):
the most efficient. But as asa side side treatments, uh, there
are definitely useful. I guess allright, thank you and art ramone rights
and the chat that he took somecolloidal silver after he got COVID, but

(01:46:06):
he can't say that it worked verymuch so silver as I understand, it
tends to have low toxicity, fairlylow side effects, but it could be
at least in certain applications of limitedeffectiveness in addressing the underlying condition as well.

(01:46:29):
But let's go to art Ramone forany closing question or comment. Yeah,
what about research into neutropics. Imean there's a lot of supplements.
I've taken a few like racitan andsome of the ones that are not quite
medication, but they're a compound thatthey make and they could a supplement legally.

(01:46:53):
So anything any interest or work ineutropics, Yeah, I mean,
as some personal interest in uh inno tropics, I've been thinking some.
I also actually have some people whocontacted me for psychoactive compounds that could be

(01:47:13):
used for many purposis mostly therapeutic,I mean that's the that's the objective.
But we have in our database,uh some some data about neuro transmitters,
some some receptors in the brain,so there could be a possibility of developing

(01:47:34):
drug that can enhance mood and maybealso improve a neuroplasticity. So far,
I've just quickly because we're we're discussingwith some people at different levels. So
here is just like some first talksabout collaboration and like looking at you at

(01:47:55):
some psychoative drugs, I see thatthat there is some interest, and we
have en up data to find thingsthat would be actually psychology. And actually
one of the things I'm taking forfor longevit is the public qunus which said
it's doper actually and it looks likeit extends the lifespan of mice quite a
lot. So definitely, probably themood is having an effect on longevits.

(01:48:20):
So the non tropic and and thepsychotropic must have some uh and some some
it and and it. It's kindof a fast a fast way to access
to access the resort. So soyes, we have some interest. There
is not our primary interest, butwe are investigating this. Yeah, I

(01:48:44):
still deal with a lot of COVIDbrain fog and I've tried various things.
Some help for a bit, butit seems like they quit working after a
while. It's like the body boatsof tolerance to it. So you know,
I don't want to keep living foreverand dealing with brain fog thousands of
years, so definitely want some neotropicto help me out. Thank you.

(01:49:05):
Yeah, we would be working onit, definitely. Yes. Actually researching
effective treatments and hopefully cures for longCOVID could be a major opportunity because a
lot of people are not only sufferingfrom long COVID, but their suffering is

(01:49:28):
quite individualized, so different people canhave very different rays of symptoms from others
who are afflicted with long COVID.I also wanted to give Alan Crowley the
opportunity to ask a final question ormake a final comment. Thank you.
I threw my two cents in onthe chat and I'll yield a ballots of

(01:49:50):
my time to the to the chair. Thank you, doctor Baber. All
right, thank you to Alan,and and I wanted to ask you,
Yvonne, what do you see asthe greatest challenges that might be standing in

(01:50:10):
the way of the realization of yourvision. What are some of the biggest
obstacles that would need to be overcomein order for your approach to become widespread
and to be used in drug discoveryessentially throughout the industry. It's really the

(01:50:32):
low base that the farmer industry istaking, so it's like it's like mammoths.
They have they're huge, but theyhave a system that's only well established
and it's difficult to make a changethere. So we really have to enter

(01:50:53):
into a door where they are investinginto collaborations, and I see there are
three four of them that are startingto build a bridge with external in the
innovative companies. And that's towards thesebridges. I think I'm going to climb
there because directly looking to some people, I mean the head of chemistry somewhere,

(01:51:18):
I mean, they have done thingsfor decades the same way, and
AI in drug discovery is a littlebit like entire aging. So people have
been selling snakeers for thousands of years, and now that we will have something
that we have already some compus thathave some effects, people still don't really

(01:51:41):
believe in them. And then withthe farmer industry, they have been told
like twenty years ago already AI isgoing to disrupt a drug discovery and nothing
really concrete happened until now, sothey're a little bit skeptical. And there
is also this problem of data haschanged. But so the challenge is really

(01:52:02):
to to talk to the right peoplethat and that to convince them to be
open minded enough to do some sometests and to change a little bit the
way they're working to optimize their theirprocess, because here we can definitely save
years and and hundreds of millions perdrugs. So it would be foolish not
not to to move forward. SoI think and I hope I'm going to

(01:52:25):
be able to to make that movementone of the main format companies soon.
But it is still a challenge becausea chemists don't come from machine learning,
and and they still think that thereis a little bit of black magic behind
the black box. So I thinkwe need to to to educate and and
to and to really enter there andand do some some proof of concept in

(01:52:50):
house with them rather than an externalproof of concepts so they can really see
how much they can save and howmuch time in in terms in terms the
feminine. Yes, it seems thatthis is an immense opportunity for those who
realize it, for those who understandthe significance of your approach in accelerating drug

(01:53:16):
discovery and the time between the initialeffort to develop an effective treatment or cure
for a disease or health condition andthe time that it's actually available to patients.
So thank you very much to doctorIvan de Weber for joining us today

(01:53:39):
and for illustrating the possibilities of machinelearning. Please go to cortexsdiscovery dot com
if you would like to learn moreabout this approach. I think it is
quite fascinating and thank you also toour panelists are USTP officers as well as

(01:54:01):
Alan Crowley. Thank you to ouraudience for the engaging questions and comments.
As always, I hope that indue time we will benefit from this approach
to discovery, and in the meantimeand for all time going forward, I

(01:54:21):
hope that we live long and prosper

All Episodes

U.S. Transhumanist Party Virtual Enlightenment Salon - Rafi Qumsieh

Episode Transcript

Popular Podcasts

24/7 News: The Latest

Dateline NBC

The Clay Travis and Buck Sexton Show

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}U.S. Transhumanist Party Virtual Enlightenment Salon - Rafi Qumsieh