George Church: Decoding the Future of Genomics

[Music] welcome to FYI the four-year Innovation podcast this show offers an intellectual discussion on technologically enabled disruption because investing in Innovation starts with understanding it to learn more visit ark-invest tocom Ark invest is a registered investment adviser focused on investing and disruptive innovation this podcast is for informational purposes only and should not be relied upon as a basis for investment decisions it does not constitute either explicitly or implicitly any provision of services or products by Ark all statements made regarding companies or Securities are strictly beliefs and points of view held by Arc or podcast guests and are not endorsements or recommendations by Arc to buy sell or hold any security clients of Arc Investment Management May maintain positions in the Securities discussed in this podcast hello I'm Nemo a multiomic Analyst at ARK invest welcome back to our FYI or for your Innovation podcast today I am joined with Professor George Church who is a well known American geneticist molecular engineer and chemist and entrepreneur uh Professor church is the professor at Harvard Medical School he's a professor of Health and Science and Technology at Harvard University at nmit and a founding member of the V Institute for biologically inspired engineering at Harvard Professor church has has many many Publications and many many patents and he's a founder of many companies in the biospace welcome Professor church so at the moment short read sequencing is dominating the sequencing space how do you think the sequencing space will evolve in the terms of the short versus the long read sequencing yeah well uh short reads are quite useful for tagging things bar barcode tagging they're good for inc2 where you're seeing where things are in living cells and living T in sorry in fixed uh tissue uh I think the long reads uh also they they tend to come from platforms that are very low entry costs so words you can get a nanopore sequence are up and running uh you know for $1,000 doll and and and uh and each run you can start getting information within an hour of of getting the sample so the this is very low barrier to entry it still is slightly more expensive and I but I think that price is going to come down um but you know with with with new uh AI machine learning algorithms it's getting more accurate I think it's it it definitely doesn't have the accuracy problem uh the long reads have have eliminated the accuracy problem of putting reads in the wrong place and getting scrambling and so forth so so now it's just a matter of improving the accuracy of you know on the individual Bas pair level do you think that we will end up using both forms or we be like a one winner and when I say the both forms do you think we'll be using kind a Bo short and and long read but for the different applications yeah I think there's definitely a need for short and long uh applications maybe sometimes even for the using them both side by side for a particular application like sequencing uh human genomes for for genetic counseling I think there are some things like repeats um that can be quite pathological and therefore you need to accurately determine them and that that requires long read when it comes to genomics data there are many concerns about the data privacy and the patient dat and how can patient data can be actually protected what do you think about the use of blockchain technology for data storage and sharing and in general how do you think about conver of the genomics and the blockchain uh Technologies well I I think there's going to be a broad range of people from those who put their genome in the public domain as as many of the people as I have and many people in my project uh and then and then there's uh We've also tried to provide the the the people that are have the most concern about privacy with with Technologies like homomorphic encryption queries and a recorded uh and those queries are recorded in in a blockchain public Ledger um so you know who asks what queries and you keep the the queries limited but the but it is possible ask things of an encrypted genome without decrypting it so that means that when when your genome is decrypted it is in principle at risk someone could go in and lift it at that moment if they can get access but if you never decrypt it and use algorithms that don't require decryption then you're a bit safer people should be aware also that your DNA can be read by anybody if you're walking around leaving your DNA on on surfaces on on uh dinner wear and cups and so forth so but I think we're entering a realm where you if you want to be very uh careful about it's it's not like it's really that dangerous um but still people should be able to get whatever they want if they want to have security I think have the technology to provide it do you think that blockchain could be a good medium for the storage of the patient genomic data I think blockchain is is a good way of of of storing the record of uh in a way that can't be forged of what's what's been done what's been sequenced who's been asking questions of it uh and even the sequence itself in principle there are plenty of ways of of encrypting genomes the thing that's tricky in storing them think is tricky is how do you ask questions of the genome without putting it temporarily how you how you can how somebody a third party can ask a question without having to receive an unencrypted form of the genome and I think that's totally feasible now very few groups are have have are using it yet but I think it it's it's definitely feasible and what do you think about DNA is a storage vehicle of the biological data in general yeah I mean it at first it sounds ironic that you're going to take your your six billion base pairs of genome compress it down to as small as three million bytes um so so six billion base pairs down to uh 3 million bytes uh and then re store that into DNA now in in a you know thousandfold compressed form that could be value I don't think we were quite at the rate rate now that we're storing pedabytes or exabytes 10 the 10 to the 18 bytes of information uh in DNA the way we are for other media there are some cases where we are storing that amount of DNA but we're not it's not we're not writing it with precise digital Precision we're not reading that amount uh but anyway the potential is definitely there it has it has long lifetime it has uh it's very cheap to make copies we've made we can make billions of copies for pennies it just is challenging to do precise writing and precise reading uh at scale um but if you want to do imprecise writing still pretty precise and and then only read the parts that you need the what you know for example a flight recorder in a uh in an airplane you don't read every flight recorder every flight every piece of it you just read the parts that are important for the rare cases where something goes wrong so I think in that model we're already ready with DNA um but it's it's getting better you know it's it's an interesting Side Market for DNA reading and writing so if I'm understanding this correctly we have to be much better in actually writing DNA and enable those Technologies given that our a given that we have a really good uh capability in reading the DNA with the sequencing right it it depends on how you're writing if you there are ways that you can write directly from the environment into DNA without going through a digital computer and that is actually much cheaper and much faster than either the electronic means or the reading methods but then now you've got now you could have exabytes of data um but but you need a way of selectively reading it because you can't read the whole thing yet there's itive ways where you can encode where things are in space and time just by where it's where it's stored in in you know let's say it's stored in an organism then or it then where in the organism it's stored is relevant to you know the illness you know so if it's sto if you have a kidney illness it'll be stored in the kidney and so that that greatly reduces the the the uh amount of reading you need to do and it and the access becomes very intuitive uh it's kind of like um ret storage and retrieval the retrieval you know where to go to get a a tiny fraction of the data but this is this is very far from routine this is just these are just academic Publications uh from my lab and some others what are your thoughts about the need and the utility of the multiomic approaches so not just kind of you know doing the whole genome sequencing but actually expanding the measurements from The genome to transcriptome epig genome proteom metabolome do you think that this kind of multiomic approach is actually providing us uh better ways to understand the biology well multi multiomics is something that's uh getting a lot of attention I think correctly it's a it's a appropriate it's because the co not only is the cost of DNA sequencing come down but the cost of RNA sequencing and tag sequencing where you can tag just about anything with a short DNA tag all of this does improve our ability to make medical decisions but this is where software does matter earlier we asked whether whether software is limiting for whole genome interpretation for you know clear uh serious melan disease and I said no but here with the multiomics there's kind of a we haven't quite caught up uh with the the and the and the problem is intrinsically more complicated integrating information about Immunology uh infectious nucleic acids um the the uh the transcript from so many different cells single cell level resolution of the of the RNA these things need to be integrated better and integrated with medical information but I think that we are severely underdiagnosed considering how cheap it is now and accurate and so the time has come to to to apply this incredibly inexpensive diagnosis to a whole variety of medical environmental and Veterinary uh uses this is the maybe also a good time to segue from the kind of multiomic um thinking to uh think how to actually hrut screening uh can be combined with the multiomic measurements and how do you think General convergence of the multiomic measurements with the highr screening uh is going to be enabled the new avenues in the drug discovery well AI machine learning is already making a big impact in in in novel protein FS rnaa drugs and delivery um Thea say protein capsids and it it will continue to expand but those are things where you have solid uh foothold and that's because uh there is so much data there's hundreds of thousands of examples of every protein category and thousands of protein categories and this helps us U design synthetic protein RNA and viral uh components that we can event test and we can test them sometimes in the tens of billions of these at once all of them designed by uh Ai and then tested and then fed back into the algorithm so so that com I think will be increasingly combined with the multiomic uh to do another level of of AI machine learning on how to use these powerful drugs in combinations for very common diseases including the most common of all which is morbidity mortality is dependent upon age um the syence uh and so that also will will be helped by the multiomics and the uh AI machine learning both for drug design and for deciding what combinations to use maybe it would be um greater to dig deeper in some of your startups from your lab which are using these approaches where they combine the kind of hrut screening with machine learning and multiomic readouts you know you have mentioned Dino but it would be great for our audience to talk about the other startups that also came from your lab in that space right so in the applications of machine learning to Medicine I say one is standing out they're they're all looking good but one is standing out already which is designing uh proteins and nucleic acids and in fact out of the 48 St in my group at least six of them have a solid leadership position on that which is Dino Therapeutics for capsids shape Therapeutics for capsids and guide rnas Jura for um IM immune um receptor uh combinations t- cell MHC um NAA for using uh antibody design and a g protein coupled receptor manifold bio is using uh designed pairs of proteins so that they can tag hundreds maybe thousands of proteins in doing clinical trials and preclinical expensive um animal trials so that uh one injection you can get thousands of results where you can tell uh which tissues it goes to the pharmacokinetics and Dynamics um you know of turnover all at once for all these proteins once so so if you really can get a thousand uh protein Pharmaceuticals um in one experiment that could be a thousand times lower costs or higher quality uh for say your expensive non-human primate trials so another question that I had for you um so the most recent research from your lab was about the Aging so maybe it's a just high level for your audience can you tell us how are you approaching the studying the diseases of Aging in your lab at the moment right so uh aging has had you a rocky start decades ago because I think there was kind of a lot of wishful thinking that it would be something simple to eat or not eat or you know exercise or not exercise and then it transitioned to no we're we're talking about very powerful medicines here um and that means you have to go through all the the complicated uh randomized clinical trials and so forth a and what's looking particularly promising are therapies because small molecules have provided most of the drugs that we love but they uh they get confused if you have complicated network with nearly identical proteins in it you know alternative splice forms um multi- Gene family members and so forth so that's one of the downsides while when you're using protein RNA gene or cell therapies you can specify exactly which family member which form you're talking about and I think Gene therapies have uh for rare diseases have have have built up this um stigma of being very expensive rightly so they are expensive and that was one of the things that worried me because when I entered the field because I like uh for things I work on to be very very affordable be equally distributed to to even the poor people in the poor Nations so what's different is with certain uh medical applications they apply to everybody so for example a pandemic is a place where it applies everybody so the the gene therapy formulations used for vaccines you know maybe 5.6 billion people uh got uh a gene therapy like formulation for the covid vaccines and the other category is uh aging uh age related diseases that's something that affects all eight billion of us and so principle the price of the vaccines was $2 to $20 and so and age related diseases could be the same or even lower such that the the government or whoever is paying for it will get more return on their investment by the by you know avoiding uh serious health consequences that are accompany age um at least putting them off all that is trillions of dollars so Gene therapies combined with common diseases like age related diseases looks promising and within the category of Gene therapies there are two classes uh I would say um all of the all the classes are multiple meaning that there's so many things you need to fix in aging that you going to need multiple drugs uh we use three or more um but then there's two classes one of which is you have to deliver it to every cell in the body that you want to fix which is not efficient yet although it's getting more efficient and the second category is things that are delivered to an easy tissue like liver and then it spreads through the blood in protein form throughout the body and uh you know I some of my companies like rejuvenate bio are P pursued both of them um using the yamanaka factors which famous for stem cell prediction but are now becoming famous again for for aging reversal rejuvenate BYO was done the first as far as I know longevity experiment in an animal preclinical uh study and that uh and that shows very significant uh longevity effects although getting it through the FDA will probably focus on um uh you know healthy you know res restoration of health and age related diseases so and then the other set so that's one set that that has to be delivered to every cell though they they don't spread so maybe some of their influence will spread but there the factors themselves the three yamanaka factors don't then the other category are some bloodborne ones from that are spread by the liver and that looks very promising as well so I think that's where I see it you know from my point of view is is the these multi- Gene therapies uh aimed at the core components of the aging process when we think about aging um you know do we think that there is like a single pathway or single like U cell type that's the kind of Key C behind the Aging or it's a more kind of mix of the things so it's the immune system it's the DNA repair system it's the kind of mitochondrial Pathways Tome it's maybe number of the viral bacterial infection during their life or do you think it's a kind of combination of all these things above that are actually uh driving the Aging so my impression uh and I think other a few other people as well is that it's all of the above is that you you might have 10 Pathways of Aging inessence cancer and so forth that you have to get them all right if you get one of them right you might extend lifespan by a year or two but it's it's it's we've gotten to the point where we know a great deal about the pathways that are involved and so we we can finally do all of them at once some some proteins will affect multiple Pathways and so you might not need 10 proteins to get 10 Hallmarks or Pathways or you could have more proteins than our pathway so you get you get redundance systems you get you know um but but anyway it's getting easier and easier to deliver multiple drugs uh multiple genes for gene therapy it's getting easier to assess the impact the cross talk between the different Pathways so I think that's that's where we're going with this so so everything on the wish list is is up for grabs all at once uh you know mitochondrial tiir you know caloric restriction type Pathways without the caloric restriction you know dealing with senescent cells either rejuvenating them or or eliminating them etc etc so I think we know we don't know everything but we know enough to start engineering you know the the history of of engineering and medical progress has been acting long before we have total information for example vaccines we had almost no knowledge of of of vir uses and uh immunology and nevertheless pretty much got the vaccination protocol right in the 1700s so I think that uh we're we're a really good place now for doing all of the Aging Pathways simultaneously in one combination therapy I would be also very curious to get your perspective on the use of the celling Gen therapies to protect us from the viral infection is that a reality you know do we think we can have like a preventive vacine that can prevent us from all kind of virus infections in life and how that would actually work I would put this in the very challenging category but we have demonstrated something like that of protection against viruses for a particular industrial microorganism we just published it this past year in nature where by change by swapping seran Lucine at at at a a particular Transfer RNA in the genetic code we can essentially the the host cell is IM is fine with it with the swap but virus is broken in every every protein of every virus and uh and we've proven that for you know a large number of of viruses probably all of them uh is is the inference now that same trick should work in every microorganism every plant every animal exactly the same uh approach uh it will be harder to do we're we're we're already have begun work on it in uh in animals especially animals will be donating organs but it it's not available yet um that's every virus and that doesn't require specific vaccinations it is essentially kind of a vaccination for all viruses at once now that doesn't protect you against pathogenic bacteria or fungi or cancer but though but there may be other uh you know out of the box ideas for those right the way it works is that you you you take the the the host let's call it you know the the the organism you care about uh which so far is eoli but in the future we'll be you know or organ donors uh and so forth and you change the genetic code which is what the there's every possible triplet of nucleotides ACG andt is in in encodes the 20 amino acids by 64 triplets so you have every possible triplets that's four to the three power 64 if you take two of those triplets because the TRNA will recognize two at once due to wobble you take you take one TRNA one two a pair of triplets and that normally encodes Serene and you make it s codes Lucine now the host is protected because you've used synonymous codons to get the host uh not using those dangerous codons those codons you're going to remap you're going to swap but now the virus is not in on that game it was it was not uh so therefore it still has the original code and the new code every place there's a Serene of this type it gets switched to Lucine so in the virus every protein depends on these searing codons and now instead of searing they're loosing which is a very different ch chemistry uh so it screws up every single protein in a way that so many things are messed up at once in one virus in each virus that they can't even evolve around it that that that's the way it looks experimentally and that's the way it looks theoretically as well we we'll we'll need to get more experience both in the industrial microorganism and also in the in the um the animals uh and human cells that we're engineering I have a followup question do you think about starting a company in this space we just we we do think that this is a valuable industrially we just published the paper um uh this this past year and so it really hasn't been a chance to talk to you about it we we want to uh I think the next thing we need to do before it becomes commercialized is to make that strain that is resistant to multiple viruses also be a strain that strain that has uh multiple uh industrial features like high protein production High you know metabolite production and so forth and so once we have those coupled then then I think it becomes a good startup uh then the next step that the the sides of going into mamalian cells that requires a considerable amount of Technology development before we even get the first Mamon cell and again it will have to be very robust and have some Cutting Edge productivity or or functionality medical functionality in addition to being M it's not sufficient for it just to be multiv virus resistant it has to also be high performance in some other category for both the industrial microorganisms and the um mamalian species uh and for the end uh what are the biggest Frontiers or emerging technology that you are developing in your lab at the moment and that you see in general in BIO research happening in next five years some examples of some emerging Technologies uh that we're very excited about in my lab in in addition to what we've already talked about is the ability to uh synthesize large genomes mamalian genomes in one go um right now we're struggling to make genomes that are a thousand times smaller than that but I think that's going to change overnight the same way that we we went from struggling to sequence single genes to routinely producing millions of whole genomes uh sequence so the synthesis of genomes is uh undergoing a revolution um our ability to do developmental biology of being able to make any cell type of being able to you know reverse aging and so forth these are all developmental things but we're going to get much better at being able to make arbitrary uh more morphologies um being able to make combinations of functionalities like different protein expression combinations um building connectomes is another developmental biology with where you can direct or enable synapse formation you know more orderly brain organoids we're working on that uh being able to read the multiomics in C2 we multiomics but in C2 where you can actually every pixel every threedimensional voxel in a in a sample uh either alive or fixed uh we can identify uh what it what its name is DNA RNA protein and possibly metabolite I think that's undergoing a revolution right now and it and all these things interface with one another where we get you know hybrid revolutions where you'll get combinations uh Professor church it was a great pleasure to talk with you and explore the all the frontiers of of innovation that you're developing in your lab thank you once again Arc believes that the information presented is accurate and was obtained from sources that Arc believes to be reliable however Arc does not guarantee the accuracy or completeness of any information and such information may be subject to change without notice from Arc historical results are not indications of future results certain of the statements contained in this podcast may be statements of future expectations and other forward-looking statements that are based on arc's current views and assumptions and involve Known Unknown risks and uncertainties that could cause actual results performance or events to differ materially from those expressed or implied in such statements [Music]

Channel: ARK Invest

Views: 5,792

Keywords: innovation, investing, cathie wood, cathy wood, kathie wood, kathy wood, cathie woods, cathy woods, kathie woods, kathy woods, ARK, market news, fintech

Id: 8Ppn5qOkf3A

Length: 30min 11sec (1811 seconds)

Published: Fri Apr 26 2024