Can A.I come to the rescue for Aria? Interview with Richard Novak of Unravel.
In our first ever podcast for the curelonp1.com youtube channel we had the great pleasure to speak with one of the founders of Unravel Bioscience based in Boston. Richard and his team are our first port of call for Aria’s drug repurposing screening using A.I. Machine Learning developed at the Wyss institute, Harvard University.
Your funds will directly facilitate this urgently needed work to screen upwards of 40, 000 molecules against Aria’s RNA to find potentially therapeutic drugs already FDA approved or even better licenced in children.
Video transcript
00:00:05.200 okay well welcome to the CureLonP1 Podcast
00:00:09.040 podcast thanks so much Chris thanks for
00:00:11.719 having me okay so we have today Richard
00:00:14.679 Novak of unravel biion
00:00:18.279 biosciences and rare shift so um Richard
00:00:22.600 we're really really excited to talk to
00:00:24.680 you um you're helping us set up this um
00:00:28.679 YouTube channel and and we really think
00:00:31.279 that obviously your your your service
00:00:33.719 and your business is incredibly
00:00:35.239 interesting to parents in our position
00:00:38.480 for parent-led research which is exactly
00:00:41.960 where we find ourselves of of parents
00:00:44.559 with an ultra rare for our daughter Arya
00:00:48.440 who has an L np1 mutation and she's the
00:00:51.879 only person in the world with this
00:00:54.520 mutation um so yeah we we wanted to have
00:00:58.399 you on because we think that it's
00:01:00.079 incredibly interesting um what you're
00:01:03.160 providing us the potential right now
00:01:05.600 that we're in our fund raise at cure
00:01:08.400 lomp one.com so and your team is part of
00:01:13.400 what we call the A Team we have people
00:01:15.759 from all over the world that we're
00:01:17.119 trying to collect and we're very proud
00:01:19.840 to have association with with such
00:01:22.000 bright minds and um really we just
00:01:24.759 wanted to bring you on so we're still
00:01:26.920 learning about unravel uh and also rare
00:01:30.240 shift and um it's an opportunity for you
00:01:33.320 to explain to people in our shoes um
00:01:36.960 tell us about the firm tell us about you
00:01:39.680 guys and why you were set up in the
00:01:41.720 first place and and and the sort of the
00:01:43.960 Genesis of of the
00:01:45.719 company yeah thanks Chris well it's it's
00:01:47.920 wonderful working with you and Andrea
00:01:49.439 and and see what all you're doing for
00:01:50.960 Arya it's it's really heartwarming it's
00:01:52.719 something that we always see with with
00:01:54.000 families you know unfortunately rare
00:01:56.079 does these families have to become the
00:01:57.439 scientists the drug developers the
00:01:59.000 experts um especially in in a case like
00:02:01.520 yours uh in Aras where there's very
00:02:04.280 little other data to to support any
00:02:07.159 understanding of a given disease and
00:02:09.038 that was sort of the motivating factor
00:02:10.840 for unravel uh for how we we started it
00:02:13.640 uh in the first place so my co-founder
00:02:15.720 Frederick and I we were both at the V
00:02:17.640 Institute at Harvard and it's a very
00:02:19.280 translational uh Institute uh we're both
00:02:22.160 on the staff side so the non-academic
00:02:24.640 side really helping commercialized
00:02:26.200 Technologies uh Drive programs out of
00:02:28.879 the lab actually so the mission is
00:02:31.000 really you can't have an impact unless
00:02:33.160 the technology leaves the lab and that
00:02:34.800 resonated beautifully with me um had a
00:02:37.480 background really uh bringing Science
00:02:39.720 Education and engineering education to
00:02:41.560 developing regions um and developing
00:02:43.920 some other solutions that were more uh
00:02:46.200 led by those experiencing the problems I
00:02:48.840 will never understand the problem but I
00:02:50.239 can uh collaborate and help uh Drive
00:02:53.040 some sort of positive impact that's the
00:02:54.840 hope at least um and so that was really
00:02:56.920 the motivating factor behind unravel it
00:02:59.360 was it was was came about because we
00:03:01.280 were working on some very challenging uh
00:03:03.159 disorders uh and disease States and they
00:03:06.640 were just extremely tough to approach
00:03:08.599 with a traditional uh biopharma mindset
00:03:11.640 typically that means start with the
00:03:13.159 target do some preclinical development
00:03:15.319 and go to the clinic and that linear
00:03:17.440 sort of reductionist approach is really
00:03:21.519 ineffective if you look at the
00:03:23.040 statistics in clinical trials that backs
00:03:24.760 me up very ineffective in actually the
00:03:27.000 very complex disorders and and certainly
00:03:29.080 lp1
00:03:30.280 um and a lot of these mitochondrial
00:03:31.400 disorders are absolutely in in that
00:03:33.840 category um and so what we did was
00:03:36.799 really come up with a with a novel way
00:03:38.720 of thinking about biology right
00:03:40.599 fundamentally we don't understand most
00:03:42.280 of it and just let's get that ego out of
00:03:44.519 the way uh we don't know that makes me
00:03:47.000 feel better because this this made me
00:03:48.720 feel very thick suddenly having to
00:03:50.760 school up on this stuff but um if you if
00:03:53.879 you're saying that then then that really
00:03:56.239 confirms the problem right is it's it
00:03:58.560 kind of just even when you If people
00:04:00.640 really understood what was happening on
00:04:02.280 inside the biology the the
00:04:04.599 bioengineering that's happening um it
00:04:08.079 just looks like somebody's come up with
00:04:09.599 some crazy ideas and if somebody told
00:04:11.840 you about it and described what was
00:04:14.200 happening in cells it sounds like
00:04:16.560 unbelievable and when you actually see
00:04:18.320 some of this modeled and actually the
00:04:20.519 cell behavior and mechanisms and
00:04:22.280 machines that are happening inside the
00:04:24.040 cell so yeah absolutely so it's
00:04:27.759 stunningly complex I mean every day they
00:04:29.840 thousands of papers that come out trying
00:04:31.240 to you know they show some sort of new
00:04:33.440 mechanism new Target new activity that
00:04:35.840 wasn't before uncovered and that's the
00:04:38.000 Fantastic interesting side of basic
00:04:39.880 science research now the challenge is if
00:04:42.160 you have a patient with a rare disease
00:04:44.160 and we work oft times with the most
00:04:45.320 challenging most severe cases where you
00:04:47.320 you typically don't have 20 or 30 years
00:04:50.000 to understand a disease and yet the
00:04:52.680 patient needs help what do you do and
00:04:55.120 that was really what gave rise to
00:04:56.600 unravel which is we found these
00:04:58.360 technology developed these techologies
00:05:00.039 at Harvard that enabled us to think
00:05:02.000 about biology in a very different way
00:05:04.000 essentially simulate what actually
00:05:06.120 happens in a patient without having to
00:05:08.680 understand it from a mechanistic
00:05:10.199 standpoint but in a way that's strong
00:05:12.759 enough to let us actually see how do we
00:05:15.080 how do we mess with it how do we perturb
00:05:17.120 it and then in obviously guide it to a
00:05:20.000 way that's favorable that restores
00:05:21.680 Health right so um in that what I just
00:05:25.479 said none of that involves knowing a
00:05:26.840 Target knowing disease-causing mechanism
00:05:30.240 um you really want to work from the
00:05:32.919 patient and then you know work your way
00:05:34.800 backwards towards a drug and then at
00:05:36.360 some point you can work towards a Target
00:05:37.880 and and essentially what we've done with
00:05:39.479 unravel and that's where the name comes
00:05:41.400 from is really unravel the disease from
00:05:43.919 a patient to a therapy and then we can
00:05:47.280 kind of fill in the the basic biology if
00:05:50.840 needed it may not even be needed and so
00:05:52.840 it's really a very different way and uh
00:05:55.280 you know it runs very much contrary to
00:05:58.160 how drugs are developed uh today today
00:06:00.039 by by big Pharma and we found that you
00:06:03.080 know we don't have a billion dollars to
00:06:04.600 throw at a at a drug with a 95% chance
00:06:07.319 of clinical failure we have to think
00:06:09.440 about it smarter and that was really the
00:06:11.080 rise of unravel biosciences now what we
00:06:14.400 were did in the process we started
00:06:16.160 working on rare diseases started with r
00:06:18.319 syndrome and that's actually um headed
00:06:20.639 to two clinical trials we just got
00:06:22.280 orphan drug designation for the drug
00:06:23.840 that we uncovered using our platform um
00:06:27.319 and there's investor interest around
00:06:28.800 that so we started on rvel right so we
00:06:30.560 spun out of Harvard about three years
00:06:31.800 ago now this extremely powerful platform
00:06:35.319 and so the whole idea was how do we
00:06:37.039 advance a therapeutic basically what our
00:06:39.919 platform does it has a deep
00:06:41.520 understanding of RNA biology of how the
00:06:43.680 genes actually interact not from a
00:06:45.560 mechanistic standpoint but if you
00:06:48.000 perturb Gene one how does it affect the
00:06:49.919 other 25,000 genes just a correlation
00:06:52.400 model you don't have to know how it
00:06:54.240 actually physically Works uh but if you
00:06:56.599 let's say the equivalent is a great
00:06:58.639 example in
00:07:00.319 traffic maps so we we like to say we
00:07:01.720 have a Google Maps for health the whole
00:07:04.639 idea is that you have a starting point
00:07:06.319 and you have a destination and you have
00:07:08.240 to get through some amount of let's say
00:07:09.599 a city in between um you know
00:07:12.879 traditional bionformatics works by
00:07:14.400 asking what's different well knowing
00:07:16.360 that you have a bed at home and not in
00:07:17.879 your office doesn't help you find the
00:07:19.440 way and you you run into these sort of
00:07:20.720 red herrings in traditional B
00:07:22.520 informatics you have all these things
00:07:23.759 that are different many of those don't
00:07:26.039 actually help you answer the question
00:07:27.520 how do you actually fix the problem so
00:07:28.720 you're refer to bi bioinformatics is
00:07:31.800 that what you just said that that's
00:07:33.599 right so so that just to Define it as
00:07:35.319 really uh you know computational
00:07:37.319 analysis of biological data uh typically
00:07:41.319 it's RNA data DNA data uh protein all
00:07:45.280 these different different bits of
00:07:46.879 information that you can collect from a
00:07:48.520 patient or some other biological living
00:07:51.360 system um the computational analysis of
00:07:54.199 that try to make sense of it make it
00:07:55.400 human interpretable is called B
00:07:57.199 informatics now the challenge is that
00:07:59.680 typically says okay you have a healthy
00:08:01.120 State and you have a disease State what
00:08:02.520 are the differences that's typically
00:08:03.800 what most of the algorithms ask there's
00:08:05.720 this whole onslaught of AI and machine
00:08:08.280 learning that gives you even more
00:08:10.319 refined differences and finds the subtle
00:08:12.120 differences between the two groups what
00:08:14.360 it doesn't do it doesn't tell you how
00:08:16.360 you actually tra travel between let's
00:08:18.280 say a disease State back to a control
00:08:20.240 State and that's exactly talk about
00:08:22.520 being I've heard you talking about being
00:08:24.720 agnostic to certain you know how how
00:08:27.639 that happened or the mechanisms
00:08:30.479 um that's that makes sense now now
00:08:32.919 you're explaining it that way so thank
00:08:35.039 you yeah and just like so and so so this
00:08:37.719 is this is AI modeling this is neural
00:08:40.200 networks we're talking here yeah a form
00:08:42.958 of it um so essentially it is a network
00:08:45.279 model um not specifically neural network
00:08:48.920 but but you know the concept is is sort
00:08:51.200 of similar everything talks to
00:08:53.399 everything else right uh just going back
00:08:56.360 to this Google Maps analogy if you have
00:08:58.760 um a card that breaks down in an
00:09:01.360 intersection obviously the cars around
00:09:03.240 it will be immediately impacted but over
00:09:05.160 time that that you know traffic jam will
00:09:07.839 propagate into the neighborhood so
00:09:09.440 everything is impacted by everything
00:09:10.880 else to some extent and essentially what
00:09:13.160 our machine learning does it finds these
00:09:15.800 these probabilities the you know what's
00:09:17.519 the fractional impact of this gene on
00:09:19.519 this other Gene um saying like if what
00:09:22.399 if a car stopped at this intersection
00:09:23.760 versus this other other intersection how
00:09:25.440 would it impact the streets two
00:09:28.079 neighborhoods away right
00:09:29.959 there's some impact it could be very
00:09:31.680 little it could be very large you don't
00:09:33.120 know until you start analyzing it and so
00:09:35.760 you know for example if a car breaks
00:09:36.920 down on a main Highway versus a you know
00:09:39.279 like a neighborhood culdesac two very
00:09:41.399 big different impacts right and so
00:09:44.160 that's what essentially our machine
00:09:45.800 learning approach studies is really
00:09:48.519 understanding those connections even
00:09:50.399 without those connections having been
00:09:52.160 necessarily um identified in the
00:09:54.440 scientific literature so unlike a lot of
00:09:56.839 machine learning companies we don't mind
00:09:59.399 the literature because it tends to be
00:10:01.240 biased based on what people were one
00:10:03.399 interested in and two got funding to
00:10:05.000 study right now you come in with a
00:10:08.000 new
00:10:09.800 St Amen to that that's a massive problem
00:10:12.519 in science and medicine as far as I'm
00:10:14.360 concerned but yeah so yeah is there
00:10:18.160 another way of sort of um deeper dive on
00:10:22.079 um kind of how that works when it comes
00:10:26.240 to so for example what we want to do is
00:10:29.880 we want to be able to work with you and
00:10:31.800 that's what our fundraising is in part
00:10:33.600 to bring you into this team because
00:10:35.120 we're so excited about working with you
00:10:37.040 on and the rare shift ability to make
00:10:39.720 that sort of accessible to us to what's
00:10:42.680 the end goal here the end goal is is
00:10:44.800 looking at something it's therapeutic
00:10:46.480 for Arya in some way the for our story
00:10:50.560 Arya is the only surviving person with
00:10:52.480 this mutation similar mutations in
00:10:55.279 children in the record died at 20 months
00:10:58.240 and so we just don't know how much time
00:11:00.120 we have with our Angel and so you guys
00:11:03.399 turning up and saying we understand that
00:11:06.160 and we think we can do something and um
00:11:09.240 obviously the power of kind of I'm
00:11:11.720 guessing Brute Force if you can just
00:11:13.519 throw CPU time or something at the
00:11:15.800 problem I don't know what the
00:11:17.360 limitations are for how that works um
00:11:21.079 but for you know explain to people I
00:11:23.680 think it would be really interesting to
00:11:25.079 understand how does this compare like
00:11:27.200 the old model might have been um drug
00:11:30.519 repurposing focused on yeast models and
00:11:32.920 that type of thing is that is that true
00:11:34.399 to say that yeah so that that falls in
00:11:37.440 the category of of uh drug development
00:11:39.440 or repurposing that's called phenotypic
00:11:41.200 screening I mean the name aside is
00:11:43.680 basically you have a living system you
00:11:46.200 make a model of your disease in that
00:11:47.959 living system and then you can basically
00:11:50.200 throw molecules at it and see which
00:11:52.079 molecule helps in some way right and you
00:11:54.279 have some some way of reading it out in
00:11:55.959 yeast for mitochondrial disorders and
00:11:58.000 and other Related Disorders
00:11:59.839 that has been fairly successful because
00:12:01.240 they have mitochondria just like humans
00:12:03.200 those the energy Powerhouse of of the
00:12:05.040 cell is pretty conserved across
00:12:07.200 organisms so they they actually are are
00:12:09.880 quite a good model system in that
00:12:12.279 context um and yeah you know the then
00:12:15.920 you have to screen thousands of
00:12:17.079 molecules in actual physical dishes in a
00:12:19.399 laboratory you use Robotics and you know
00:12:22.399 it might take you a few weeks few months
00:12:24.040 to screen a few thousand molecules
00:12:26.160 essentially but the upside of it you are
00:12:28.320 working with a real living system no
00:12:30.800 question right even the parts you don't
00:12:33.199 understand are in that system right so
00:12:35.160 that's the upside now it's not a
00:12:37.440 vertebrate it's not uh it's not Arya so
00:12:40.279 there you always have to deal with you
00:12:41.440 know the fact that it's a model system
00:12:43.199 versus an actual patient you obviously
00:12:44.839 can't we understand even we understand
00:12:47.519 even I think we've got a delay on the
00:12:49.680 line but forgive me for cutting across
00:12:52.160 even with critically on the lp1 in a
00:12:55.360 yeast model there is some differences I
00:12:57.800 think in the genes in the yeast
00:12:59.959 expression of the lp1 versus human
00:13:03.160 expression that's right that's right
00:13:05.079 there's some differences um some
00:13:06.880 differences are known some are unknown
00:13:08.880 right and and so you get back into the
00:13:10.880 question how much biology do we
00:13:12.560 understand is this the correct model or
00:13:15.160 a correct model a valid model uh given
00:13:18.000 the context you don't know until you go
00:13:21.160 from that model to a patient and then
00:13:23.320 you you know the reality check comes in
00:13:26.240 I I I guess I'm interested how how you
00:13:28.800 Cali calibrated in the first place how
00:13:30.639 do you calibrate those tools or you your
00:13:32.880 your platform as you refer to I don't
00:13:34.399 know if it's given a name or a secret
00:13:37.320 name or a name that you refer to your
00:13:39.760 platform as but how is that was that
00:13:42.360 calibrated against things like yeast or
00:13:44.639 real
00:13:45.920 organisms so it's a human model so we
00:13:49.000 the platform is called bav so that's
00:13:50.680 really the computational uh you know
00:13:52.800 drug and Target Discovery platform um
00:13:55.880 essentially what it does the input is
00:13:57.759 RNA sequencing data so looking at how
00:13:59.880 the genes are transcribed going from DNA
00:14:02.480 to RNA and that's sort of one of the you
00:14:04.800 know key parts of of living living cells
00:14:07.839 living organisms and then we take that
00:14:09.880 information we have an algorithm that
00:14:12.079 converts it from this list of genes and
00:14:14.440 how much of each gene is in in the
00:14:16.399 tissue being collected and converts it
00:14:18.360 into essentially this Google Maps
00:14:19.920 traffic map where are the traffic jams
00:14:22.360 what's going on with that patient and we
00:14:24.680 compare the patient data relative to a
00:14:28.120 um healthy
00:14:29.720 usually so in this case you you would
00:14:32.399 sample Arya and you would sample arya's
00:14:34.759 Mom that's right that's right that's
00:14:36.800 typically a good scenario or oftentimes
00:14:39.360 uh oftentimes it's the parents uh
00:14:41.839 sometimes they're siblings of the same
00:14:43.160 sex and actually that matters quite a
00:14:44.880 bit um the idea is that we're trying to
00:14:46.839 see okay how do we make the the patient
00:14:49.759 as close to whatever that that healthy
00:14:52.440 control you know first-degree relative
00:14:54.079 so a lot of the genes are the same a lot
00:14:56.040 of the environmental factors are the
00:14:57.639 same same time zone typical similar
00:15:00.720 sleep schedules diets all of that plays
00:15:03.680 into health and how people respond to
00:15:06.399 drugs the response to um you know
00:15:09.440 genetic mutations other factors all have
00:15:12.160 an impact um and the idea is that by
00:15:15.680 looking at the RNA we capture all that
00:15:18.199 so it goes beyond DNA analysis so DNA is
00:15:20.759 typically how patients get diagnosed
00:15:22.480 because there's a mutation you can do
00:15:24.600 DNA analysis is actually very robust
00:15:26.519 because DNA doesn't change except I
00:15:28.720 guess cancers but in the case of these
00:15:31.199 uh genetic disorders the DNA is not
00:15:33.680 changing but the disease is right and
00:15:37.319 it's not because of the DNA alone the
00:15:39.079 DNA is encoding certain genes and has
00:15:41.160 certain predispositions but then it has
00:15:43.360 to interact with environment um the the
00:15:46.279 in this case you know Arya is growing
00:15:48.440 developing um that all factors into how
00:15:52.279 the disease progresses doesn't or
00:15:54.199 doesn't progress and and the
00:15:57.360 bit the bit way you could help me out
00:15:59.880 was just understand from a technical
00:16:02.000 point of view how there's an RNA is
00:16:04.880 produced from the the lumpy one we
00:16:07.600 focused on the whole system and I'm
00:16:11.399 guessing all the genes that are being
00:16:13.279 expressed in those samples um so that
00:16:17.120 would mean there's lots of RNA right and
00:16:19.040 I'm guess I'm just curious about how
00:16:20.680 much RNA and how you look at all of the
00:16:22.560 RNA
00:16:23.720 together right so from a diagnostic
00:16:26.759 standpoint what Arya uh received as the
00:16:29.279 diagnosis was that laan P1 is mutated
00:16:32.639 right um and that obviously is is the
00:16:34.800 root cause of this disorder that's that
00:16:37.480 part was clear from the diagnosis it
00:16:38.959 does not tell you how to treat it um and
00:16:42.000 so that's sort of where we come in and
00:16:43.920 and it doesn't even tell you necessarily
00:16:45.360 how the disease will progress or or
00:16:47.399 other aspects um so there's this type of
00:16:50.120 analysis that often times a lot of
00:16:51.519 Foundations do which is called genotype
00:16:53.319 phenotype analysis so genotype is the
00:16:55.680 DNA how where is this Gene mutated let's
00:16:58.440 say Where's La lawn P1 mutated for each
00:17:01.319 individual patient and then correlating
00:17:03.440 that with symptoms let's say seizure
00:17:05.839 developmental delays cognitive function
00:17:08.319 Etc um that relationship can be
00:17:10.839 correlated right now sometimes that's
00:17:13.959 quite promising but many times there are
00:17:16.160 a lot of uh sort of uh confusing results
00:17:20.240 we sometimes actually now we work with
00:17:22.280 one uh Foundation now there two girls
00:17:26.039 completely different families they have
00:17:28.119 identical mutations in the particular
00:17:30.559 Gene of interest um they have are on the
00:17:33.240 complete opposite spectrum of the
00:17:34.840 severity of the disease right clearly
00:17:37.559 that Gene alone does not explain what is
00:17:40.080 actually happening um in those patients
00:17:42.720 it's absolutely the gene causing the
00:17:44.600 problem that's not that's out that's not
00:17:46.240 the the question the question is what
00:17:48.200 else is involved and essentially what
00:17:49.799 the RNA does it it you mentioned there's
00:17:52.240 a lot of genes there's about 25,000
00:17:54.120 genes in in the human genome give or
00:17:55.840 take it turns out they all have these
00:17:57.880 variations they're these inherited
00:17:59.600 factors they spontaneous variations
00:18:01.919 mutations um that are not causing
00:18:03.960 disease but they can certainly affect
00:18:05.919 the outcome of drug treatment the
00:18:08.720 disease itself the the the root cause of
00:18:10.880 the disease and a bunch of other factors
00:18:13.440 um it could be you know how quickly you
00:18:14.880 metabolize a drug I mean that's been
00:18:16.840 studied very very deeply because it
00:18:18.919 impacts every drug you take how how much
00:18:20.520 you metabolize it uh but there also some
00:18:23.000 mutations that can actually help uh
00:18:24.880 counteract the disease in a patient this
00:18:26.880 has been studied for example in
00:18:27.919 Parkinson's and Al patients um where you
00:18:30.919 might have two siblings even even U you
00:18:32.799 know fraternal twins or somebody very
00:18:34.520 very closely genetically related one
00:18:36.520 develops the disease let's say 30 years
00:18:38.039 before the other what's going on it's
00:18:40.000 not it's not the the the what you think
00:18:42.159 is the causitive gene I mean that may be
00:18:44.080 causing the disease but that's not the
00:18:45.840 full story and essentially that's where
00:18:47.240 unravel bav platform comes in it's not
00:18:50.240 about that one intersection of the car
00:18:51.799 being you know broken down uh it's the
00:18:54.240 whole city map that you care about and
00:18:56.120 how do you actually travel from a
00:18:57.360 disease State back to a a healthy State
00:19:00.480 and essentially that's what our machine
00:19:01.720 learning platform has identified now
00:19:04.320 when it comes back to patients like Arya
00:19:07.400 traditional AI approaches require
00:19:09.679 tremendous amounts of data right I mean
00:19:11.840 just staggering amounts of of data what
00:19:14.360 we've done that's quite unique is we've
00:19:16.039 essentially you asked how how we sort of
00:19:19.039 U you know calibrated our model well so
00:19:20.799 our model is actually human data that's
00:19:22.600 driving it entirely so we can feed in
00:19:24.919 data from other species uh but
00:19:27.080 everything is processed in the cont
00:19:28.720 context of a human and so we've looked
00:19:30.919 at you know the 25,000 or so human genes
00:19:34.480 and we've analyzed if you perturb one
00:19:36.960 gene what effect does it have on all the
00:19:38.600 other genes and we obviously went
00:19:40.520 through that iteratively and then we
00:19:42.120 essentially that's how we built up the
00:19:43.320 model now that took a massive amount of
00:19:45.799 computation but it's done wow so now
00:19:48.480 what it enables is that we can take
00:19:51.000 essentially uh arya's data arya's RNA
00:19:55.760 data and interpret that in the context
00:20:02.799 for some reason I just lost your audio
00:20:08.320 briefly okay all right well I think
00:20:10.440 we're back now hopefully um so yeah
00:20:12.880 hopefully this will keep saving it 20 or
00:20:15.240 so seconds
00:20:17.799 so um I don't know what's quite the
00:20:20.799 issue there but I I can't also see you
00:20:23.280 um
00:20:27.400 recording hm
00:20:29.400 yeah I I think I got you back I think I
00:20:31.559 got you back okay great it should be
00:20:35.480 fine for the
00:20:36.600 recording yeah sure we'll just have a
00:20:39.120 cut there I'm sure but um thank you um
00:20:42.840 so that was really really interesting as
00:20:44.640 well you're in full flow so I would love
00:20:46.880 to hear what you what I
00:20:49.559 missed yeah so so the whole idea is
00:20:52.240 right that that we looked at um all this
00:20:56.520 this staggering amount of data from
00:20:59.039 human tissues and where essentially one
00:21:01.600 gene was perturbed and we looked at what
00:21:03.360 does it du to all the other 25,000 genes
00:21:06.200 and we went through that and this is
00:21:07.440 where we built up this very deep uh
00:21:09.840 machine learned model of human health
00:21:12.159 right and that took a tremendous amount
00:21:14.640 of of data processing now what that
00:21:16.919 means is that and we did the same thing
00:21:18.840 for about 40,000 existing molecules
00:21:21.400 instead of a single Gene perturbation
00:21:23.080 now add aspirin add Tylenol add you know
00:21:27.679 a Statin um and ask no other information
00:21:30.880 or no other question other than what is
00:21:32.840 it h what effect is it having in this um
00:21:35.320 Network state so now we have this really
00:21:38.039 interesting scenario where we don't need
00:21:41.159 a huge amount of patient data we can
00:21:44.320 start with a single patient and a single
00:21:46.559 matched control and we can start asking
00:21:48.880 the question how do we convert that you
00:21:52.919 patient into uh the healthy control and
00:21:56.400 essentially what we do we run simulation
00:21:59.000 we import patient data uh which actually
00:22:01.919 I can talk about in a moment but we
00:22:03.960 import patient RNA sequencing data um
00:22:06.720 into our platform and then we can
00:22:08.279 essentially simulate what happens if you
00:22:10.679 give someone aspirin what happens if you
00:22:12.559 give someone this particular Statin and
00:22:14.880 then we can see where does that model
00:22:17.799 tell us the patient would go most drugs
00:22:20.600 are irrelevant they just don't do
00:22:22.080 anything as we know also clinically
00:22:24.000 that's that's true some drugs um may
00:22:27.120 make the disease even worse right they
00:22:28.799 may may be something you don't want to
00:22:29.919 take other molecules and these are the
00:22:31.559 ones that we're looking for are which
00:22:33.000 ones actually go from let's say
00:22:35.200 arya's state that we we have interpreted
00:22:38.640 um to where her mother is right so the
00:22:41.799 idea is to to to drive that path from a
00:22:44.760 disease state to a healthy State um and
00:22:47.640 essentially we run um you know
00:22:49.559 contrasting that with the phenotypic
00:22:51.080 screening that we discussed earlier we
00:22:53.159 can run 40,000 drugs in a few minutes on
00:22:56.760 each patient on each patient sample and
00:22:58.679 so we can even look at over uh drugs
00:23:00.679 over time drugs at night drugs during
00:23:03.320 the day which ones could actually be
00:23:05.400 therapeutic um in that patient's
00:23:07.720 particular
00:23:10.120 context exactly so we have a human model
00:23:13.559 of of Health uh we have a a database
00:23:15.960 about 40,000 U molecules including a lot
00:23:18.760 of the repurposable approved molecules
00:23:21.240 for for some indication and then we can
00:23:23.880 essentially go to Just individual
00:23:25.600 patients and really we are very much
00:23:27.480 patient left patient Centric U because
00:23:30.799 we see every patient as an individual uh
00:23:33.960 you know patient or disease independent
00:23:36.000 of how they were diagnosed and so what
00:23:38.039 we do is we essentially import the
00:23:39.720 patients's RNA profile into our database
00:23:42.320 or into our bav uh platform and then we
00:23:46.080 essentially run uh if you think of it as
00:23:48.520 uh you know a n of one or single patient
00:23:51.320 clinical trial on 40,000 molecules one
00:23:53.880 at a time and let's just see in the
00:23:55.760 computer what happens that's all we do
00:23:58.080 right I mean that's all we do it's a
00:23:59.720 complex process behind the scenes but
00:24:01.880 it's a very simple question that we're
00:24:03.640 asking right which molecules help the
00:24:06.159 patient which ones are neutral and which
00:24:08.320 ones might actually make disease worse
00:24:10.600 so I could visualize it I could
00:24:12.279 visualize it like some kind of like a a
00:24:15.320 graph or some kind of state of this is
00:24:18.240 this is dysfunction this is disease and
00:24:20.760 this is the control this is health and
00:24:23.919 your in my layman's terms would be
00:24:26.840 understanding does this molecule move us
00:24:29.039 closer or further away from that is that
00:24:31.640 so I'm guessing that's right you're
00:24:33.200 doing that's right
00:24:35.720 um okay and and so from an RNA point of
00:24:39.200 view um are we looking at only the RNA
00:24:43.640 that comes from the L P1 or are we
00:24:46.120 looking at all of the RNA that would be
00:24:48.799 expressed in the is it a nasal swab or a
00:24:53.000 cheek swab yeah so let me let me talk
00:24:55.520 about that for a second so of course
00:24:57.360 when you have a computation platform the
00:24:59.360 patients not digital so you have to
00:25:02.240 somehow you know digitize them or
00:25:04.200 incorporate them into the platform and
00:25:05.960 so the way we found out we can how we
00:25:08.320 can work is through nasal swabs so
00:25:10.960 thanks to this lovely covid pandemic uh
00:25:13.000 everybody knows how to swab and we
00:25:15.880 actually found a way to uh instead of
00:25:18.760 looking at the pathogen side the virus
00:25:20.720 or or bacterial uh side of of the
00:25:23.240 equation you when you do a nasal swap
00:25:25.399 you scrape up some of your own cells the
00:25:27.240 human cells what we were able to do is
00:25:29.520 show that from just the few cells that
00:25:31.720 you scrape up and it's very complex in
00:25:33.760 inside your nose you know your your
00:25:35.320 blood vessels close to the surface
00:25:36.720 immune cells um epithelial cells sort of
00:25:40.000 like skin but but you know sort of
00:25:41.559 almost like inside of your lung is
00:25:42.919 actually what you're uh inside of your
00:25:44.919 nose looks like um we basically take all
00:25:47.919 that information collect the cells and
00:25:50.000 and convert it into RNA data so called
00:25:52.440 RNA sequencing and this really Builds on
00:25:54.960 the personal Genome Project over the
00:25:56.640 last you know 253 years
00:25:59.120 um all the technologies that were built
00:26:00.559 up including the ones that were used
00:26:01.880 often times for the DNA diagnosis that
00:26:04.520 that Arya received for example um but
00:26:07.120 instead of the DNA we look at the RNA
00:26:08.919 but to your point it's not just lawn P1
00:26:11.520 um this is completely as as you
00:26:12.880 mentioned the term agnostic we don't
00:26:15.320 know what to look for so let's look at
00:26:16.840 everything and that's really we don't
00:26:18.679 know what we don't know and that's
00:26:19.919 really the guiding philosophy yeah only
00:26:22.159 thing we know is that the disease is in
00:26:24.000 that patient and that patient's disease
00:26:25.640 is different from potentially a
00:26:26.919 different patient we make no asss about
00:26:29.039 any of that it sort of seems to really
00:26:31.559 embrace the complexity and and remove
00:26:34.080 assumptions as a model absolutely these
00:26:37.080 diseases are are staggeringly complex
00:26:39.600 and and so we don't know what we don't
00:26:41.640 know so let's just take as much data as
00:26:43.720 we can and learn from it have the data
00:26:46.000 tell us what we should be looking
00:26:48.279 at so the specific thing about the L P1
00:26:52.399 expression and the RNA versus all the
00:26:55.159 RNA that might be expressed in those
00:26:56.760 cells help me on that
00:26:59.880 yeah so L P1 is the cause of the of the
00:27:02.919 disease but what we're interested in is
00:27:05.320 how to fix it so it's a slightly
00:27:07.080 different question than a diagnostic
00:27:09.200 question so we're not on a diagnostic
00:27:10.840 Odyssey we're on a therapeutic
00:27:12.799 Odyssey what what's very unique about
00:27:15.279 unravel is that we make the uh the
00:27:18.080 hypothesis that how you treat a disease
00:27:21.440 does not require reversing the the
00:27:25.120 disease itself how it was caused there
00:27:27.279 may be other me ganisms and there's a
00:27:29.039 lot of data out there in many other
00:27:31.000 diseases um where that's true so it
00:27:34.080 gives you a lot of flexibility on and
00:27:35.960 how you could potentially uh treat the
00:27:38.120 disease biological systems are extremely
00:27:41.399 uh so-called plastic so if you have some
00:27:43.720 sort of U stimulus whether an injury or
00:27:46.360 something you know extra nutrients the
00:27:48.360 body adapts um it you know it doesn't
00:27:50.880 need to be this very consistent system
00:27:52.519 and this is this is because of evolution
00:27:55.000 the organisms that were able to adapt
00:27:56.760 survived right that's really the nature
00:27:58.720 of it so yeah humans as an organism are
00:28:02.039 very able to adapt including at the cell
00:28:04.240 level so what we do is see okay the the
00:28:07.200 mutation in Lan P1 in arya's case causes
00:28:10.240 the disease let us have the software
00:28:13.360 model how do we fix it now one mechanism
00:28:16.279 could be by focusing on something
00:28:18.039 directly linked to lwn P1 in this case
00:28:20.960 um you know clearing out bad proteins in
00:28:23.399 mitochondria that's sort of the the
00:28:24.720 function of lawn P1 maybe it's somehow
00:28:26.440 disrupted it might have other function
00:28:28.080 fun that have never been been documented
00:28:30.679 right for all we know that happens all
00:28:32.279 the time um our platform lets us work
00:28:35.440 with that sort of uncertainty or or
00:28:37.840 unknown um and just simulate okay if you
00:28:40.159 add a drug right what does it do does it
00:28:42.720 fix things uh again from this this
00:28:44.760 biological Network standpoint so we look
00:28:46.399 at all the and how are you measuring how
00:28:48.320 are you measuring the fix as it were
00:28:51.360 what what is the because you're not
00:28:53.600 looking at the RNA at that point you're
00:28:55.320 looking at something else right that
00:28:56.880 says so we still look at the RNA the
00:28:59.880 outcome okay yeah so we still look at
00:29:02.080 the RNA essentially sort of um think of
00:29:04.919 it as an as an as a network state right
00:29:08.440 um like a pattern that you're going
00:29:10.279 after so there's there's uh in this case
00:29:12.519 arya's mother's RNA pattern and that
00:29:15.200 we're defining as healthy this is where
00:29:16.760 we're trying to get to and then there's
00:29:18.799 ARS pattern overall will be very similar
00:29:20.840 but there G be some some areas where
00:29:23.000 it's not the same so it could you are
00:29:25.799 you saying it's Ultra focused on simply
00:29:28.399 the RNA and that's how that's
00:29:31.000 essentially what we would assume if we
00:29:33.760 get to correcting the RNA expression is
00:29:38.240 that what was saying right so the the
00:29:41.679 even though we're reading out only the
00:29:43.159 RNA again it's a biological system so
00:29:45.799 DNA gives rise to RNA RNA gives rise to
00:29:48.200 proteins they're all these metabolites
00:29:50.320 modifications it's a very complex it's
00:29:52.200 not a nice line that we might have
00:29:54.279 learned early in in school everything's
00:29:56.840 talking to everything else but by using
00:29:59.360 RNA it's a very robust way of collecting
00:30:01.640 information on this whole complex system
00:30:04.200 yes you absolutely have gaps we are not
00:30:06.039 reading out the protein we're not
00:30:07.360 reading out metabolize we're not reading
00:30:09.000 out modifications to the proteins a
00:30:11.120 bunch of other factors but at some point
00:30:13.640 it feeds back into the RNA and
00:30:15.799 essentially that's sort of like with
00:30:17.320 Google Maps you don't have to track
00:30:19.159 every single car on the highway to have
00:30:20.720 a sense of where is there a traffic jam
00:30:23.159 if you track every 50th car you probably
00:30:25.440 have a pretty good idea right and this
00:30:27.240 the same thing with RNA it's it's uh
00:30:29.799 it's obviously imperfect but it's good
00:30:31.519 enough to let us see what's going on at
00:30:33.480 this really complex system level and
00:30:36.440 that's really how we work um and so it's
00:30:38.519 it's all the genes they come from from
00:30:40.960 the swab to our surprise that actually
00:30:42.760 gives us really good correlation with
00:30:44.440 what's happening in the brain and and
00:30:45.840 other tissues um and we've been able to
00:30:48.760 use this for other patients where
00:30:50.519 basically the the top drug that we
00:30:52.200 identifi essentially this this you know
00:30:53.799 in silico you know computational
00:30:55.760 clinical trial that we run we took the
00:30:58.320 the drug that was computationally
00:30:59.559 predicted to be most helpful there there
00:31:02.399 were a couple families that started
00:31:03.840 taking it uh for two different diseases
00:31:06.039 and have had some really profound
00:31:07.960 positive effects right so the idea of
00:31:10.120 repurposing can be actually very very uh
00:31:12.799 therapeutic it's not sort of just
00:31:14.919 historically it's been focused on
00:31:16.000 mitigating symptoms so for example if
00:31:17.799 you have seizures how can you prevent
00:31:19.840 seizures if you have you know GI
00:31:22.440 problems or cognitive problems let's put
00:31:24.120 that aside and let's let's do something
00:31:25.440 else later that's historically how
00:31:28.000 purposing was was thought of or even
00:31:29.480 just drug development let's let's come
00:31:30.840 with a better seizure medication what we
00:31:33.720 do thanks to looking at it from a
00:31:35.360 network standpoint the the hypothesis
00:31:37.919 which so far has been panning out quite
00:31:39.639 nicely um even in in patients is if you
00:31:42.880 can fix that RNA Network right the one
00:31:45.120 that we've been talking about this kind
00:31:46.000 of computational network of of patient
00:31:48.480 disease State and reverted to that of a
00:31:50.120 control you fix a lot of problems at
00:31:52.840 once so essentially it's it's
00:31:54.799 counteracting sort of that the cause of
00:31:56.960 the disease or the impact of the disease
00:31:59.360 in some unknown way right or sometimes
00:32:02.039 known sometimes unknown
00:32:04.480 yeah how it approaches I'm still trying
00:32:06.679 to get my head around it even though um
00:32:09.200 but it sounds it sounds my kind of um
00:32:12.399 way of approaching problems and I think
00:32:14.519 just incredible how how people have
00:32:16.840 figured this out um but I think it
00:32:19.399 sounds you know you've got to you've got
00:32:22.080 to harness nature right you've got to
00:32:24.559 say you got to respect it and remove all
00:32:27.279 the assumptions and so for me working
00:32:29.639 and solving that problem in that way is
00:32:31.679 is is working with that elegance and
00:32:34.080 that beauty and and accepting that we
00:32:36.799 you know we there's so much that we
00:32:38.440 don't know um but I even said even
00:32:41.639 though you've been working with me so
00:32:42.960 well trying to explain it it's still
00:32:45.039 kind of um you hard for me to understand
00:32:47.960 and I I wonder for other viewers um as
00:32:50.760 well for them to take away but I think
00:32:53.039 for parents in our shoes um you know
00:32:56.440 give them you gives people hope it gives
00:32:58.960 people tremendous hope you know it's
00:33:01.440 definitely um uh you know light in in
00:33:04.679 the darkness for us and our family um
00:33:07.399 that we're you know doing this fund
00:33:09.320 raise and we kind of implore people to
00:33:11.840 sort of check out um unravel and to
00:33:15.600 check out our website to Cump one.com
00:33:19.279 and hopefully donate um and if you can't
00:33:22.120 donate share and so that we kind of want
00:33:25.880 um you know people to be aware of of
00:33:29.200 what you're doing um unravel biosciences
00:33:32.240 and also rare shift so probably give um
00:33:35.159 give people a little taste on what's the
00:33:37.039 difference between the two what what
00:33:39.559 rare shift is versus unravel biosciences
00:33:42.639 and also we've talked about in silico
00:33:45.440 which is super interesting what about in
00:33:48.159 Vivo um an animal study to move things
00:33:52.279 forwards and help people in our shoes
00:33:54.720 parents who might be out there looking
00:33:57.200 for help
00:33:58.320 um how should they understand and
00:33:59.960 interpret
00:34:01.360 this yeah thanks Chris so really we have
00:34:04.919 sort two arms we have a very innovative
00:34:06.919 business model and the whole idea is
00:34:09.280 that we um have this very powerful
00:34:11.359 platform that we can use um internally
00:34:14.320 as a biotech company which is what we
00:34:15.760 are we're a Therapeutics company based
00:34:17.599 in Boston sort of one of the major
00:34:19.599 biotech hubs um and our goal is to
00:34:21.639 develop drugs right and and eventually
00:34:23.440 sell them and through and develop them
00:34:25.760 to through the clinic and into the
00:34:27.040 market so that's really what we are as
00:34:28.520 unravel now what has become very
00:34:31.239 apparent there's this massive unmet need
00:34:33.639 in neurodevelopmental or just genetic
00:34:36.159 disorders um usually neurod Deval
00:34:38.960 because that's been sort of the biggest
00:34:40.639 Gap historically but fam started coming
00:34:43.480 to us essentially wanting repurpose
00:34:45.679 drugs because they heard that we use
00:34:48.040 existing molecules for discovering new
00:34:50.119 mechanisms new targets right and so for
00:34:52.760 for us what's a discovery tool is
00:34:55.040 potentially either a temporary or
00:34:56.639 potentially the end treat treatment for
00:34:58.560 an actual patient and so over the last
00:35:01.079 year year and a half we actually started
00:35:03.040 partnering with families letting them
00:35:04.800 use through a nonprofit model nonprofit
00:35:07.200 service model our platform and so our
00:35:09.359 platform consists of this insilico model
00:35:11.119 of human health which is absolutely
00:35:12.440 revolutionary and where in a matter of a
00:35:14.480 month or two we can get to a list of uh
00:35:17.599 you know existing molecules that you can
00:35:19.320 work with your clinician to take off
00:35:20.839 label we cannot encourage that we we we
00:35:23.400 uh you know that's entirely a personal
00:35:25.160 decision um but we can certainly
00:35:28.079 generate the data for that um but then
00:35:30.680 from a longer term standpoint we can go
00:35:33.040 all the way to clinical trials either
00:35:34.480 with existing drugs or use that
00:35:35.920 information and develop newer uh
00:35:37.720 potentially even better molecules that
00:35:39.119 are now tailor made for the therapies
00:35:41.359 and this really addresses a a major un
00:35:43.359 unmet need in the rare disease space um
00:35:45.359 because if you have let's say one
00:35:46.520 patient um no Pharma will touch that
00:35:49.359 that's that's you know personalized
00:35:51.079 medicine is not a good business model
00:35:52.960 because you only have one exactly and
00:35:54.520 this is a fundamental problem for people
00:35:56.400 in our shoes you know and this is we're
00:35:59.079 completely alone without people like you
00:36:01.560 guys and so what we did very
00:36:03.960 intentionally was open up this platform
00:36:05.720 just through a nonprofit model so it's
00:36:07.839 not free but it's just the bare minimum
00:36:09.720 cost to operate it um and families and
00:36:13.680 Foundations get repurposing um and we
00:36:16.079 get data and and the potential path for
00:36:17.960 commercial development right and so it's
00:36:19.599 sort of a win-win uh where we charge the
00:36:22.000 minimum but but then we can be uh your
00:36:24.160 partners in drug development for both
00:36:26.200 existing drugs so this could be your
00:36:28.839 traditional repurposing path we can make
00:36:30.560 we're also doing this making new
00:36:31.920 formulations synthesizing drugs if
00:36:33.839 they're not readily available as
00:36:35.359 generics um Case by case we can do all
00:36:38.520 kinds of things um our chief medical
00:36:40.440 officer has brought eight drugs to
00:36:41.599 Market so uh you know as a company we've
00:36:43.800 done this many times um but also being
00:36:47.160 the partner all the way through new drug
00:36:49.280 development kind of building on the data
00:36:51.640 including clinical data oftentimes
00:36:53.760 showing hey this path will work this
00:36:55.440 target that we've uncovered that our
00:36:57.079 computation tells us is really really
00:36:59.160 you know interesting
00:37:00.520 therapeutically this will work let's
00:37:02.720 build a better molecule in one case
00:37:04.720 we're doing this for red syndrome we're
00:37:06.440 using an existing oncology drug but we
00:37:08.720 found a new Target and it turns out it's
00:37:10.839 metabolite right so it goes through your
00:37:12.560 liver gets degraded but actually that
00:37:14.599 degradation makes an even better
00:37:16.280 molecule for for red syndrome and so
00:37:18.480 we're actually commercializing that as a
00:37:20.160 new molecule that will of course take
00:37:22.079 years of development but in the meantime
00:37:24.440 we're running two clinical trials to
00:37:26.359 show that this existing molecule could
00:37:28.160 be one an immediate
00:37:30.640 treatment and two showing that this
00:37:33.440 mechanism that we've identified actually
00:37:35.040 works in patients right so how how do
00:37:37.520 you think that could work for Arya in
00:37:39.280 that in that sense what do you predict
00:37:41.440 you know once we could hopefully get
00:37:43.240 this going what would that look like do
00:37:45.359 you think for Arya given what you know
00:37:47.920 about lp1 or um you know this is useful
00:37:52.800 to know personally I'm kind of
00:37:55.319 intrigued yeah so really we get the list
00:37:57.720 of drugs out right so these are
00:37:59.040 basically a personalized list of drugs
00:38:00.599 for ra and we always say that the
00:38:02.720 disease Discovery and and and
00:38:04.440 therapeutic Discovery has to happen at
00:38:06.240 the individual patient level there is no
00:38:08.319 average magic patient it is Arya it is
00:38:12.079 Billy it is whoever right there they
00:38:15.000 have their unique combination of genes
00:38:16.680 environment and even just the
00:38:17.880 progression in time for their disease
00:38:20.680 that makes the disease very unique now
00:38:23.040 we basically take it on ourselves as
00:38:24.480 unravel is to figure out okay we can do
00:38:26.839 Discovery at a personalized level how do
00:38:29.319 we actually make this make commercial
00:38:30.680 sense and that's really the transition
00:38:31.920 from rare shift which is patient by
00:38:33.560 patient nonprofit model to there may be
00:38:36.920 a business opportunity here right and if
00:38:40.000 there is everybody wins it's it's you
00:38:41.599 know the idea is that that that feeds
00:38:42.880 back into the the patients as well but
00:38:46.160 it is up to us to essentially find
00:38:47.560 connections across these rare ultra rare
00:38:50.240 and even n of one disorders and form
00:38:53.119 basically commercial opportunities and
00:38:55.079 that's an unravel problem but we're
00:38:56.800 basically trying to make sure that at
00:38:58.119 least one thing that you guys as parents
00:38:59.839 don't have to deal with is the business
00:39:01.359 side of things you're already scientists
00:39:03.640 you're already caregivers you have all
00:39:05.119 these other responsibilities uh let's
00:39:07.359 not add you know biotch executive to to
00:39:09.560 that as well um and so the whole idea is
00:39:11.920 that unravel sort of serves this bridge
00:39:13.839 between you know to some extent kind of
00:39:15.680 a nonprofit uh early Discovery often
00:39:18.160 times you know aligned with academic
00:39:19.760 Discovery but then seamlessly
00:39:21.440 transitions into uh you know biofarma
00:39:24.200 and and including 505 B2 reformulation
00:39:27.400 at least with the US FDA um but all the
00:39:30.040 way to new drug development sometimes
00:39:32.000 you have to develop a new molecule an
00:39:33.480 existing molecule may not accompl what
00:39:35.599 you want so to answer your question we
00:39:37.839 don't know but we will find out
00:39:39.720 relatively fast what that looks like
00:39:41.599 right and so what kind of what kind of
00:39:43.400 timelines would you be thinking so
00:39:45.839 typically it takes about a couple of
00:39:47.280 months um unless there's a you know
00:39:49.960 urgent need to accelerate things to work
00:39:51.960 with the patients get their RNA sequence
00:39:54.200 run through a platform now you also
00:39:56.359 alluded to the fact that we also work
00:39:58.160 with preclinical models so we actually
00:39:59.960 use um Tad poles so your xenopus leis uh
00:40:03.800 you know African claw frog uh embryos we
00:40:06.839 genetically engineer them with let's say
00:40:08.599 in this case lawn P1 to make little
00:40:11.800 patient avatars of uh of whoever uh it
00:40:15.960 is of their particular disorders could
00:40:17.520 be one or more genes and then we can
00:40:20.160 actually screen the molecule so that's
00:40:22.040 typically how we progress so we go from
00:40:23.760 patient to a drug list look at the drugs
00:40:26.280 look at the ones that make the most most
00:40:27.520 sense for translating go back into these
00:40:30.839 tadpole models and take us a couple
00:40:32.319 weeks to make a tadpole model so it's
00:40:34.200 actually really really fast much faster
00:40:36.079 than mice even faster than than um like
00:40:38.480 even fruit flies um and then lets us
00:40:41.000 work very quickly but importantly this
00:40:42.960 is why we like whole animals whole R
00:40:44.960 birt animals because we can ask about
00:40:47.400 neuromuscular function seizures growth
00:40:50.480 and development sleep they sleep at
00:40:53.000 night like humans do right all these
00:40:54.760 factors even some social interactions
00:40:56.480 autism pheny
00:40:57.839 we see that and we have this wonderful
00:41:00.240 uh computer vision algorithm that we
00:41:02.000 that we developed we basically through a
00:41:03.839 camera we can analyze all these metrics
00:41:05.599 for thousands of animals at once right
00:41:07.960 so that lets us actually essentially
00:41:09.599 take Arya which which is one patient and
00:41:12.880 you want to be of course very careful
00:41:14.800 Safety First in terms of anything that
00:41:16.680 actually ends up in Arya but this is
00:41:18.599 where you go into preclinical models use
00:41:20.480 the animal model test all kinds of
00:41:22.960 hypothesis all kinds of drugs drug
00:41:24.640 combinations do the target Discovery and
00:41:27.359 we can drive that very quickly so in a
00:41:29.000 matter of months we can get to let's say
00:41:31.560 a therapeutic mechanism have a molecule
00:41:33.960 that can engage it ideally it's a
00:41:35.160 molecule that can be consumed by a
00:41:37.200 patient already so it's already approv
00:41:39.520 and then we can move towards a clinical
00:41:40.800 trial so we've done this now several
00:41:42.200 times where um in always under a year
00:41:45.800 but uh you know now as as low as about
00:41:47.599 two to three months we can go from
00:41:49.560 working with a patient developing a
00:41:50.920 tadpole model um and showing that
00:41:53.000 potentially if the patient and the
00:41:54.520 family decides with their Clin to do
00:41:56.040 something off label we can get clinical
00:41:58.079 at least proof of concept very quickly
00:42:00.240 okay we can do very rapid proof of
00:42:02.200 concept clinical trials under the
00:42:04.000 appropriate regulatory uh guidance but
00:42:07.040 then we also show that it works in these
00:42:08.760 tadpole models oftentimes in par to
00:42:10.920 developing mice or other more more I was
00:42:13.800 thinking that in in ARA's case would we
00:42:16.160 then look at Mouse models too and um
00:42:19.400 this type of thing now that that's
00:42:21.720 really really interesting and um I
00:42:24.079 appreciate you running through that it's
00:42:26.240 um you know the the gene itself um in
00:42:29.359 terms of you know taking it you know to
00:42:33.240 a state where it can impact other people
00:42:35.480 and there certainly we hope that when
00:42:37.960 people are donating to help us do this
00:42:41.400 work we're looking primarily to extend
00:42:43.720 arya's life so that we can um you know
00:42:47.720 work on additional longer term
00:42:50.119 treatments for Arya but the gene in
00:42:52.359 question lp1 increasingly is of interest
00:42:56.680 from um as I understand it things such
00:42:59.720 as Cancers and you will know more than
00:43:03.240 than me clearly on its impact but if if
00:43:05.559 you were to develop a therapeutic and
00:43:07.599 took it to that point where you could
00:43:10.280 you know produce something on scale that
00:43:13.160 has an impact on who knows things like
00:43:15.680 kodas syndrome or um you know
00:43:19.200 potentially Cancer Treatments or
00:43:21.319 something it's difficult to predict
00:43:23.319 isn't it but it's certainly an an
00:43:25.839 Endeavor that we see AC across the other
00:43:27.920 partners that we're working with and
00:43:30.160 we're so pleased that that people are
00:43:32.559 taking an interest um because of that
00:43:35.240 but but it to me it seems that there
00:43:37.160 could be um a potential to really help
00:43:40.000 other people is really what I'm driving
00:43:41.720 at absolutely I mean that that's
00:43:44.280 something that that we routinely see
00:43:45.680 something that we strive for as a
00:43:46.880 company as well is find these links
00:43:48.839 between uh disorders or or other disease
00:43:51.640 indications that may not have been
00:43:53.240 uncovered before because that's how you
00:43:54.760 have the biggest impact uh in the most
00:43:56.520 efficient way
00:43:58.359 it's so interesting and I'm sure you
00:44:00.599 know uh we can talk again I'd love to do
00:44:03.640 that um from from my side just wanted to
00:44:06.680 thank you again for spending the time
00:44:09.280 with us today um I think you've got a
00:44:11.520 heat wave over there in Boston maybe at
00:44:13.640 the moment um somebody told me um thank
00:44:17.240 you very much for coming on and and
00:44:19.520 doing with us uh today I really
00:44:21.800 appreciate it um and i' just like to
00:44:24.240 thank you so much for um yeah doing what
00:44:27.920 you've done being so supportive
00:44:30.480 throughout and we're just really looking
00:44:32.480 forward to the opportunity to to do this
00:44:35.079 type of work
00:44:36.319 together well thanks so much Chris
00:44:38.280 thanks for the opportunity it really
00:44:39.599 really pleasure to speak with you and
00:44:41.000 and just I think you know together we
00:44:42.559 can really change how we go about
00:44:44.160 treating diseases let's hope so and and
00:44:46.880 we're really really fingers crossed um
00:44:49.280 going to make a difference for ARA and
00:44:51.680 um we're just really super stoked to to
00:44:54.920 have this opportunity to learn about it
00:44:57.119 it we want other families to be aware of
00:44:59.119 it and um you know this patient Le space
00:45:03.160 um or parent Le uh space is sort of I
00:45:06.440 suppose increasingly a thing and you
00:45:09.839 don't want it to be a thing and I
00:45:11.720 suppose culturally we found that in the
00:45:13.880 states um people are more open to that
00:45:16.160 than perhaps is the case here in the UK
00:45:20.079 um but it we're just very grateful and
00:45:23.480 um again on a family level every body um
00:45:28.319 who knows Arya is is super grateful that
00:45:30.640 this is on the horizon for us and so um
00:45:34.280 we'd love to um you know speak again
00:45:36.920 sometime soon but thanks for today
00:45:38.640 appreciate it thanks so much Chris
00:45:41.840 [Music]