Startup Battlefield Competition: Session 3 – Genesis Therapeutics

Hey everyone so I'm Andres Papito and I'm going to talk a little bit about Drug Discovery and how we're planning to Make it better So the reason that drug Discovery is Hard is that even once you've figured Out all the biology you know why a Particular disease happens Um what proteins you need to bind to to Fix the disease there's still a lot of Work and that happens because you need To in general find some molecule that Can act as a drug that will have the Desired effect and at the same time Respect a lot of the other properties That a drug needs to have like not being Toxic getting to the right part of the Body and so on and traditionally how This is done is with uh very smart by Very smart people with phds called Medicinal chemists and they have to do It with a combination of basically their Their expertise and intuition and A lot of trial and error and The reason that there's so much trial And error and the reason that it's so Expensive is that there's no there's Really no way to know short of actually Trying it out whether a uh how a Particular molecule is going to behave So Genesis comes from uh our co-founder Evan's PhD work at Stanford in Vijay Panda's lab which you may know because It's also the birthplace of folding at

Home and Uh yeah and Evan published a paper Called potential which is kind of the Foundation of our AI platform Which was basically field leading in Predicting molecular properties So briefly Obviously Predicting molecular properties via Computer is very desirable people have Been trying for a long time people have Even been trying with ML or AI methods For a long time and so I'll give an Example of how Our our current Technology based on this Potential at paper Uh in one way is a big improvement over The things that came before and so here We can see that Before the kind of wave of deep neural Networks we had hard-coded features you Might have a chemist say or this Particular type of ring or this Particular particular type of ring is Something to look out for and then you Draw on kind of old school uh ml Techniques against that like svms or Random forests more uh then once the Kind of AI wave hit we kind of shifted Gears and now we have people trying to Kind of cut and paste techniques from More developed Fields like image Recognition and basically try to treat a Molecule as kind of a 3D image where

Instead of these Square pixels you have Kind of these Cube boxes called voxels But that has some problems like for Example Uh you no longer have explicit Representations of the bonds that Connect atoms and turns out those are Very important or there's a lot of empty Space in this representation so you Waste a lot of computation processing All this empty space and so this kind of Like copy pasted approach from a Different field it doesn't actually work So well and where the field is now is More things that are kind of Directly directed at uh predicting Chemical properties And for example uh one feature potential Is that it uses as its representation a Graph which is you know pretty Straightforward but this wasn't what was Being done for a long time and so in the Graph you have the nodes of the graph or The atoms of that molecule the edges are The bonds and you can kind of calculate Things from first principles and it Solves a lot of those problems So in addition to our AI platform we Also have a couple other like very key Uh competencies in particular I want to Call out our Chemical expertise so we think we can do Great things with AI we still think that Developing drugs is an incredibly

Complicated process and so we so we have Strong experienced people on the team Who have basically done this all before They've produced multiple FDA approved Drugs drugs and clinical trials all Without computers and for the most part And so we're very excited to see what They can do with our technology And Again so one of the problems is Basically that as you're developing a Drug you need it to be effective but you Also need all of these other things to Line up and this is really hard with as Basically a Human chemist because you have Effectively this kind of very high Dimensional space of chemical matter you Need to choose which parts of it to look At and investigate and spend your Resources on and it's kind of very ad Hoc and it's at human Pace right so it Would be much better if we had accurate Predictions of what a particular Molecule in there is going to do if we Could if we could explore the space more Systematically automatically kind of Sweeping through it at high speed with The computer So here's our team we'll call it in Particular Pepe and pepe process and Nick stock our uh our chemistry experts And uh yeah and Evan of course who is uh Whose research at uh Stanford this is

All based on And yeah and briefly kind of why are we Doing this of course uh you know biotech And developing drugs is perfectly viable As business model it's also Uh very uh very key to us uh to everyone That's on the team so far that uh that This is a process that's going to make The drug the drug development industry More effective and basically help people Get medicines for diseases that Currently they don't have very many Options for And yeah ready to take questions All right judges So Go ahead yeah yeah so this is this is Really cool this is really awesome Audrey uh what kind of diseases are you Guys starting with like are you are you Going to focus on first yeah so we have A lot of flexibility because the Platform and our our Medicinal Chemistry Team are not specific really to any Particular area so it's going to depend On Um on what we think is promising in our Kind of like initial surveys of various Things we can work on as well as if we Partner with for example large Pharma Companies it will depend on what targets They're interested in and they may have Existing programs in various stages of Development but in your current kind of

Surveys and Analysis what have you found To be the the one area that you are Pretty comfortable that you have a you Know a pretty high confidence interval Of yeah Um we haven't decided on a particular Initial set of diseases yet okay Um it's basically so far we've been Working more on developing that core Technology Uh and and working on uh you know Finding deals with farming companies What exactly is the business model Yeah so basically uh our our kind of What we can do is we can make this Process more efficient and the ways that Translates to a business model is either That we would develop uh develop drugs For diseases kind of In-House up to some You know up to some point kind of the Further you develop them the more Clinical trials you do you know the the Better it is but you can also hand them Off at various points to some of the More established players and the other Way is to do something directly in Collaboration with a larger player So he's the point to create a discovery Workflow tool or an analytics platform That you sell into Pharma and cros or to Develop your own lab Uh yeah either can work we're not in Particular we're not especially Innovating on the like the lab the

Physical lab component and obviously Despite that can be more efficient to Have our own at some point when we when We get big enough at the moment it's Mostly cro focused It's a super competitive space already With obviously like gtn and lab genius Are you after small molecules how do you And how do you fit into that landscape What's your USP yeah so uh yeah we're we Are looking at mostly small molecules Um In general we're not too worried about The crowdedness of the space because you Know there's A huge amount of untreated uh untreated Or disease or things that need medicine Where there's no good medicine available So we think there's basically plenty of Plenty of room And in also we have a couple things that We think make us stand out and the two I Would call it in particular which I did Actually mention a little bit is One this AI platform which is built on This kind of field leading peer-reviewed Research uh and our CEO is the person Who kind of invented that technology And the other aspect is uh those highly Experienced medicinal chemists that kind Of work very closely with us both using The the results of our of our platform To be more effective you know what can They do when they have access to these

High quality predictions as well as Feeding back into it as we like tailor The platform for specific targets and Programs but you haven't set up a lab Yet Wet lab We are in the process of starting to do Tests you don't need to build your own Lab to do that but if you want to build Your targets in-house is that not what You're going to need to do Um you can get pretty far with with These contract research organizations uh You know our expertise and our strong Point is basically in choosing what Chemical matter is interesting and what We think is promising not necessarily Stitching those atoms together Which is a little more mechanical Hear a little bit more about why this Problem is emotionally resonant with the Team like what's your superhero origin Story we heard a lot about the science But yeah Founders quitting that so often Kills startups yeah Um Yeah well I'm kind of the lucky one but Both of our Founders have some amount of Like experience with uh you know a Physical condition that there's not Great medicine for so it's pretty Personal uh in that regard to some of Our team And in particular people at the top

Thanks The original research it all came out of Stanford right did I hear that right Yeah during uh Evan's PhD PhD research In Vijay Panda's lab okay And just to fill in the sense of What's Um you said the the AI platform was part Part of the secret sauce how does it Compare to to sort of any sort of Generic horizontal AI platform that you Might use in in the Pharma vertical What's special about it yeah Um Basically the tools that large farm Companies are using are not nearly as Cutting Edge you know they don't have The people working at them that are on The Cutting Edge inventing these things Right you know there's only one Evan He's working at Genesis Um Yeah okay Final question One more do you plan to move out of Small molecules Uh well I think there's there's a lot of Promise and probably plenty of work for The near future uh You know as we prove our prove our Tech On more and more things we can you know Address more and more fields or more Ambitious projects but you know lots of Things can be treated with small Molecules so there's not kind of any

Immediate pressure What sort of the Milestones you're Hoping to Sort of achieve in the next quarter or Two quarters how are we going to how do You prove that this it works the model Is going to be effective et cetera Yeah So basically uh it kind of parallels the Process of drug Discovery at a whole Which is there's kind of a uh just in General there's kind of a lot of Checkpoints of you know does it does it Bind to the Target does it have this Particular you know does it kill people Does it do this does it do this and kind Of as you progress you address more and More of those things and so we'd want to See progress along that timeline like Present in our results as we test things Okay one more round of applause for Genesis Therapeutics [Applause]