welcome to the Department of Electrical
and Computer Engineering distinguished lecture series this lecture is also
co-sponsored by the university committee on lectures we are very privileged and
honored to have dr. Lera hood with us today I guess the best introduction that
I can give to him is that we had to delay the presentation for 10 minutes to
put more chairs to accommodate the people who are flowing in and I I just
watched the number of chairs in the room double over the last last 10 minutes dr.
Leroy hood is the president and co-founder of the Institute for systems
biology in the past he served as the William Gates chair of the department of
molecular biotechnology at the University of Washington he made many
outstanding contributions throughout his career in biotechnology systems biology
medicine and bringing engineering to medicine he led the biotechnology
revolution from the front by inventing five pivotal instruments related to DNA
RNA and protein sequencing including the DNA sequencer which enabled the human
genome project he cured the first neurological disease by gene transfer in
mice and he played a role in founding more than 14 biotechnology companies
including such highly successful and large companies as hanjan and applied
Biosystems dr. hood has published over 680 papers received 26 patents and 17
honorary degrees he co-authored several textbooks in biochemistry immunology
molecular biology and genetics and is currently finishing up a book on systems
biology he co-authored a popular book titled the code of codes on the human
genome project it’s on sale in that corner although I haven’t asked dr. hood
for permission I’m sure we can trick him into autographing books if you if you
like he’s one of only seven scientists in the
United States to be simultaneously elected to three National Academies the
National Academy of Science the National Academy of Engineering and the Institute
of Medicine dr. hood is currently focusing his energies on creating a
paradigm shift in medicine to what he calls p4 medicine predictive preventive
personalized and participate Rhee he recently co-founded a company called
integrated Diagnostics to create a platform company for this paradigm shift
and he will be talking to us about it today and before I forget this stock
will be followed by a reception so please stay back ask questions and
you’ll have an opportunity to interact with dr. hood please join me in
welcoming dr. hood tires and kudos well thank you for that gracious
introduction I’d like to take you on a whirlwind tour
of where I think meta medicine is going to be going in the future and I’ll talk
about it’s the origins of the change the technologies of the change the role
systems biology has played in the change but most especially I want to get across
the idea that this new medicine is really predicated on the idea that
medicine is an informational science and to give you an example my prediction is
in ten years or so each individual patient will be surrounded by a virtual
data cloud of millions of data points and will have the IT for healthcare to
reduce that enormous data dimensionality two simple hypotheses about health and
disease so what I’d like to do then is talk a little bit about the framework
for this revolution the idea of complexity and the paradigm changes have
led to it I’d like to talk about the fundamental pillars of p4 medicine and
then really what p4 medicine is is societal implications and then how are
we actually going to bring p4 medicine to patients what are going to be the
major challenges when I was thinking about the 21st century 10 years ago 15
years ago it was clear to me at that time that the fundamental challenge for
all of these scientific and engineering disciplines was the challenge of
complexity and what I became convinced of is that biology is uniquely
positioned to deal with its complexity because of this idea of biology and
medicine being informational in nature systems approaches emerging technologies
and and new analytic tools what’s important
about this supposition is this gives a biology the wherewithal to attack some
of the most fundamental problems in society
not just healthcare which I’m going to talk about but global health the
environment energy nutrition agriculture and the like so I’ll argue that the
paradigm changes and approaches I’ve laid out here could be applied equally
well to any of those other areas as we move ahead when I was a young assistant
professor at Cal Tech back in 1970 or so I read this book by Thomas Kuhn on the
structure of scientific revolutions and it’s really about paradigm changes and
it’s about arguing he discussed what paradigm changes were and how they were
brought about and the reason I became so fascinated is it was clear to me and it
turned out to be true that I wanted to participate in a number of different
paradigm changes and I’ll tell you about four paradigm changes I’ve participated
in that really have fundamentally laid the framework for the fifth paradigm
change which we’ll talk about today which is p4 medicine
so the first paradigm change had to do with bringing engineering to biology and
that was the development of lots of technologies that you’ve already heard
about the second came about is a consequence of developing the automated
DNA sequencer namely in the spring of 1985 I was invited to the first meeting
that was ever held on the human genome project and it became obvious to me at
that time that this was a transformational and revolutionary
opportunity for biology what came from the santa cruz meeting where 12 Siot 12
relevant scientists were asked to come and talk about the genome project were
two interesting facts one that it was technically possible although at that
time certainly difficult and two there was an enormous split 50 to 50 and a 50%
that were again the genome project were really vehement
Lee against it and in fact when you went out in the community at that time the
the split was probably ninety to ten at least initially and one of the most
dominant organizations that was against the genome project until the very end
was the National Institutes of Health something that people don’t remember
National Academy Committee I was a member of that had half opponents and
half proponents that she gave a unanimous endorsement in 1988 and that
really turned the tide in NIH quickly saw that this was an opportunity rather
than a than a challenge that came from developing the automated sequencer was
the realization that we had to bring into biology all sorts of different
kinds of scientists and I went to the chair of the president of Caltech in
about 1988 87 and I proposed I create a cross-disciplinary biology department
there and the biologists vehement Lee opposed it and in 92 it was possible
with Bill Gates helped to actually move to University of Washington and set up
the first cross-disciplinary department and this department was incredibly
successful it invented the first fundamental technologies of proteomics
yet pioneered the software that fueled the genome project with regard to
assembly and and quality verification we there invented the inkjet printer that
is what Agilent uses today to make DNA arrays and synthesize long DNA strands
and so forth we had a whole series of different things but it was obvious even
at that time that what I really wanted to do was in fact systems biology and
what I became convinced of is the bureaucratic constraints of a State
University made it impossible to do many of the things that were required to
really realize systems biology so I resigned from the University in 2000 and
started this Institute for systems biology and we’ll talk a little bit more
about that in just a few moments so it was the integration and synthesis
of these paradigm changes that led to the the final paradigm change that is
the subject of this lecture now I would make a number of observations about
these paradigm changes so observation number one is that they really
fundamentally change tell either we thought about biology or certainly how
we did biology observation number two was it’s unbelievable how conservative
most scientists are and how much difficulty they have in thinking outside
the box and this is the topic I’m going to return to at the end of my lecture
and number three was the really interesting idea that for every one of
these ideas we had to create new organizations in order to realize them
and again we’ll talk more about that at the end of the lecture so the four
fundamental pillars of systems biology p4 medicine or exactly what we
enumerated earlier the idea that biology is an informational science that the
system’s approaches mandate that we approach the complexity of biology in a
holistic manner that we have to invent new technologies that will let us
explore new dimensions of patient data space and finally we need the analytic
tools to be able to handle all of this information so what I’d like to do is
talk about each of the four of these and give you an idea of how we’ve moved
forward either in thinking about these and or in developing new approaches
and/or technologies to mediate them so let’s begin by talking about biology as
an informational science and I would argue there are really five fundamental
ideas that you have to understand in this context if you’re really to be able
to do systems biology so idea number one is the idea that evolution is not
directed rather it builds on random mutations and in
mental conditions to shape unique solutions to various types of problems
and the really key point is it doesn’t use the law of Occam’s razor it’s much
more it’s much more evolution in X and a Rube Goldberg apparatus because
evolution builds upon what went before and in fact the well-known fact they’re
a series of insects called the stick insects that evolved indi evolved wings
three or four different times and with each of those events there was recorded
in the genome information and it led to a group Goldberg like kind of machine so
for those of you who are engineers and computer scientists and theoretical
types the only way to de convolute the complexity of biology is to start with
the data itself that is the only way we’re going to gain deep insights into
the into the data second fundamental point is modularity is really the key to
being able to understand complexity and to scale the different levels of
biological complexity so all of this is obviously made by legos and you can see
different different entities here that have very different functions and once
you understand the basic lick legos you can then use them to build structures
once you understand the structures then you know what all the related structures
might be doing and so it goes on with biology there are fundamental modules
that can be used in many many different ways that solve fundamental information
processing problems and by coming to understand these fundamental modules
again it’s a real key to D convoluting complexity there are fundamentally two
types of biological information there’s the digital information of the genome
there are the environmental signals had come from outside the genome
it’s the collision of these two types of information that creates the three
processes of life or for evolution development physiologic response and
disease and we have to understand these processes then in the context of what
are the relative contributions of these two fundamental types of information a
second point that is really important is how do you connect those two types of
information with the phenotype the appearance of the organism and we do
that by two types of informational structures one are biological networks
which capture and transmit and integrate in mind and ultimately pass off
information to the second type of structure complex and simple molecular
machines and the interface between these two is not always obvious but these lie
at the heart of what systems biology is understanding their dynamics both from a
temporal point of view and from a spatial point of view is absolutely
critical now to give you an idea about relative contributions to phenotype of
environment and genome these are the left index finger prints of two
identical twins and you can see they’re absolutely a different from one another
so this means if they have the same genes and they do more or less that the
differences are environmental and the question that is absolutely fascinating
is one the the differences rise as a consequence of stochastic events that
occurred during the differentiation process of the fingerprints or do they
occur as a result of stochastic events that come in from the environment and
modify the output of the fingerprint program and everything we don’t really
know the answer to that and the final point I would make about information is
that it is multi scale it is hierarchical and it is absolutely
critical if you’re to understand a particular system on one level say
the single cell and how it divides you have to capture all of the antecedent
levels of information right back to the genome because the environment in pages
upon and modifies every single one of these levels of information so we have
to be able to capture the information at levels and carry out an integration
process with which operationally explicate s–
the environmental contributions at each level to the digital signal and we don’t
know how to do this very well at all we’re still in the process of learning
it’s a it’s a terrific problem to think about for students and so forth so these
are these five fundamental principles and either argue
they are absolutely essential to really thinking in a deep way about systems
biology okay let’s talk about systems approaches to biology and what that
actually means so by view of systems biology and I’m
going to give you a simple analogy because I really got invited by an
engineering department so my question is suppose you were an engineer and you
wanted to figure out how a radio converted radio waves into sound waves
the first thing you do obviously is you take it apart and you’d look at its
individual components and maybe come to understand what those individual
components could do and this of course is what biology has done primarily for
the past 40 years or so looking at individual genes and individual proteins
what the genome project brought in the way of a revolution was a complete parks
list and for the first time we could do coherent systems approaches to the
understanding of biological complexity the second thing the engineer would do
obviously is assemble those parts back into their circuits and then come to
understand individually and collectively how these these circuits converted
electromagnetic waves to two sound waves and that’s exactly what we
and biology the environment in digital genome of impinge on human and
biological circuitry gives a phenotypic response normal and/or disease and if
you think about systems biology I would say they’re really fundamentally several
really important components number one is the idea that you take all of the
information about a complex system you have and you formulate a model which can
be descriptive graphical and/or mathematical if you have enough
information and then you formulate hypotheses to test that model and you
explicate those hypotheses by perturbations of the system you
integrate the data from the systems perturbation back into the model
changing the model in accordance with the reality that’s been revealed by
those data and you do this in an iterative fashion until you bring into
proximity experimental data with the model itself and the ultimate idea of
what we want to be able to do is to generate models that are predictive and
actionable so actionable means you can actually tell a patient’s something that
will be useful for them to do and improving their health and we’ll talk a
lot about that a little bit later on systems biology has a very rigid set of
views of with regard to data number one it argues wherever possible you should
do global data acquisition because you never know which data is really
important we can do that in genomics we can’t do it proteomics and metabolomics
yet although in proteomics we’re beginning to get close you have to be
able to integrate the different data types I’ve already talked about that you
have to be able to delineate the network dynamics molecular machine dynamics and
again do it in both temporal and spatial sense and finally and this is really key
with large data sets most of what you look that varies normal to experiment
is noise how do you deal with the enormous noise in large datasets
one of the biggest fundamental problems in biology and I’ll tell you there are
two types of noise one type is technical and the other is biological because if
you measure any aspect of phenotype often it is the sum of several different
biological phenomena and if you’re only interested in one of them you have to
subtract out the others so how do we do that I’ll tell you one way we can go
about doing that later on and the final point of course is discovery science
that is being able to numerate all the objects in a biological entity so you
have a complete parts list as absolutely key to systems biology we’re struggling
to do that now with with protein proteomics and metabolomics and so forth
what we’ve really predicated our approach to systems biology on is the
idea that biology is the driver of Technology in that in turn as the driver
of computational and mathematical tools and it have the pace has to be set by
the biology and we’ve created a real cross-disciplinary culture and there are
two things you have to be able to do with cross-disciplinary cultures one
you’ve got to teach people to learn to speak the languages of the other
disciplines and two you have to learn to work together effectively in teams to
take on big problems and actually three you have to learn how to educate new
students to be cross-disciplinary and that’s going to be really important in
the future and if you can do that with that philosophy you can generate them
information that’s useful in the context of the cross-disciplinary environment
you see there on the lower right-hand corner of this slide what we’ve done it
is B as well as we’ve created a lot of high-throughput facilities that any
scientist has access to as well as the bioinformatics and computational
resources to be able to analyze that information and then apply it to simple
and complex model systems and ultimately to human organs
so ISP is now 10 years old it has 11 faculty and about 300 staff and a
digression for just a moment it’s always interesting in 10th anniversary to think
about metrics of success and we actually have two really interesting metrics of
success let me say when I started the Institute of systems biology some 10
years ago there was an enormous skepticism and and the proclamation that
systems biology was all real life and what has happened it’s been very
interesting is last year a report came out from the National Academy called the
new biology for the 21st century and it’s a perfect description of the
systems biology that I’ve told you about here and in fact it argues that this is
going to be the future of biology and medicine and what’s interesting is two
of my friends that were the biggest skeptics about systems biology wrote
this report we’re on the report so people who have come around to a
different way of thinking I asked one of them I said you know why you call it the
new biology which is terribly vague why didn’t you call it systems biology and
they said we didn’t want to give the Institute too much credit so anyway now
the second metric is a personal one and for us it’s been absolutely spectacular
and it arose as a consequence of a Spanish Institute that determined a new
way of assessing impact factors on individual papers and then assessing
over a five-year period all of the papers that a particular Institute had
created and then giving an average impact factor for that time and they
looked at 2,200 institutes and these are in all fields of science and by that
measure the Institute came out first in the US and third in the world in terms
of the impact of its papers and one of the really important observations was
that we had high-impact papers in biology and medicine in computation in
mathematics and technology so it it belies the power
it underlies the power underlines the power of these these systems approaches
now a systems view of disease in this network centric
view of systems biology I’ve given you means that in the relevant diseased
organ one or more networks becomes diseased perturbed either in terms of
genetics or in terms and or in terms of environment that authors the structure
of the network that alters the envelope of information that it dynamically
expresses and the argument is that dynamically changed envelope of
information explains the pathophysiology of the disease and gives us brand new
insights into diagnosis into therapy and ultimately into prevention and I’m going
to show you with one example that all of this is true and the example we’re going
to talk about is a neurodegenerative disease in mice prion infection that
we’ve studied for the past seven or eight years together with George Carlson
at the McLaughlin Institute and the essence of this disease is absolutely
fascinating it arises from a protein a normal protein who can have its
three-dimensional structure altered to become quote infectious once the
infectious protein arises it has two interesting properties one it can
catalyze the conversion of the normal to the prion infectious form so the disease
is autocatalytic and this is actually one of the most dangerous infectious
diseases it’s the disease that there no neurosurgeon ever wants to touch and
usually they throw the tools away they don’t even try and sterilize them and
the second thing is that that infectious protein is utterly resistant to all
known protease and almost certainly that’s a part of the basis for the
reason it aggregates the neuronal cells you kill something leads to the you
degenerative phenomena but for us the reason that it was a nice model system
is if you infected normal animals with these infectious particles you could
time precisely the beginning of the disease process and follow it through to
its termination so you could do dynamical studies all the way across
disease progression and that’s exactly what we did with several inbred strains
of mice the idea was to infect one strain and then to look at normal
counterparts we analyzed the brain transcriptomes using DNA array
technology that let us visualize most of the genes and the most important thing
that we did was that each of the 10 time points or so subtract the disease
transcriptome from the normal transcriptome to get genes that differ
and of course we were horrified to find that initially 7400 genes different
that’s more than 1/4 of the most genes differed one from another so the idea
was that there is enormous noise in those 7400 genes and the
question is how do you get rid of it so what we did was to infect 10 different
prion strain inbred strain combinations that each exhibited different aspects of
biology that we could subtract away from the fundamental biology we were
interested in namely the genes that encoded this neurodegenerative prion
response and to give you an example one of the animals we looked at was a double
knockout for the prion gene it never got the disease when infected with
infectious prions so any changes in its brain transcriptome could be subtracted
away they were irrelevant to the disease process and so we were able to with
those eight subtractions get down to about 300 differentially expressed genes
so that was more than a 20-fold increase in signal-to-noise so the really
interesting question is what do we do with those 20 gene at those 300 genes
now and the idea was to take the four most
fundamental processes in the end prion disease and to look at the networks to
the extent we knew them and then to map onto each of those four networks those
333 genes that actually changed across the 10 time points in the progression of
the disease so we looked at these differentially expressed things for all
four of these networks and made a series of really key observation so the first
observation that was fascinating is the four major networks were sequential e
disease perturbed in an order that never changed prion accumulation first and
then glial activation and then two different forms of neuronal degeneration
so it means there’s the sequential process and if you’re interested in
therapy you want to focus it on using drugs that can make the initial Network
early on behave in a more normal fashion to abrogate the entire disease process
and this is the strategy that will revolutionize how drug targets are
chosen and we’re beginning to test this in yeast as a model organism at this
point in time the second thing is if you looked at the dynamics of how this is
the prion accumulation Network actually changed you could come to a number of
absolutely fascinating conclusions and remember we looked not only at this
network but at three other major networks and as you’ll see in just a
moment we looked at some other ones so the major conclusions were one we can
say about 300 genes were involved in this process
number two we could say about two-thirds of those genes mapped into the four
major networks but the other third mapped into six other networks that
nobody had any idea were involved in this disease process and with all of
them we can see how it made a lot of sense for them to be involved finally as
I’ve already mentioned the disease the networks for disease perturbed in a
sequential fashion but the really important point is
collectively these network changes explain every aspect of the
pathophysiology of this disease that we knew for the first time ever it also
opened up new ways of thinking about drug target discovery as I already in
human numerated and it got us thinking about a completely new approach to
diagnostics and approach where we will make blood a window into viewing health
and disease for the individual and the approach that we took to doing that was
really twofold one was to look at those networks and get key changed nodal
points and ask were they secreted into the blood and if
so were there concentrations altered in accordance with their transcript
concentrations and with many that we looked at the answer to that was yes so
that led us for the first time do beautiful pre-symptomatic diagnosis of a
disease long before any clinical science ever showed up in these animals and I
think the second approach we’re taking is going to turn out to be maybe the
most important and that is we could use deep transcriptome analysis of 40 organs
and humans and mice to identify transcripts that were uniquely or
relatively uniquely expressed in each organ and then ask were those secreted
into the blood and of course if they were secreted into the blood then that
meant that created a fingerprint of say in the case of the blue brain specific
proteins and the idea is that those fingerprints and the blood then will
record reflections from the disease process
that occurs in the brain and each of the proteins in the brain specific
fingerprint and we’ve now identified for human and mice more than a hundred such
proteins okay each of those proteins reflect the operation of a particular
cognate network so if the brain is normal the hundred levels will be set at
1:1 thermostat stage and if there is a disease process a few of the disease
perturb networks will modify the reflections of their cognate proteins
and every disease as employees different combinations of disease perturb networks
so for each different disease you get a unique fingerprint in the blood
hence you can use this to distinguish health from disease and if you have a
disease you can distinguish which type of disease it actually is and we’ve used
these markers to do pre-symptomatic diagnosis to stratify disease that is as
you know most diseases such as breast cancer aren’t one disease there are
probably five or six or seven diseases and we’ve shown that we for some of
these diseases can actually do the stratification from the blood we can
actually in the case of prion disease beautifully follow the progression of
the disease showing markers coming up from
sequentially the four different networks and then we can look at the response to
therapy and recurrences and in fact as you heard we started a year and a half
ago a company called integrated diagnosed Diagnostics that using exactly
these strategies in connection with targeted mass spectrometry and the
results really have been absolutely spectacular to date so what about
emerging technologies how are they going to increase the patient data space
dimensionality that we can begin the search and let me talk about four big
science driven technology driven projects and I’ll talk a little bit
about three of them but not about fourth so one is the sequencing of families
which uniquely marriage marries for the first time genetics in genomics and
gives us enormous power in a finding disease genes and I’ll explain that in
just a moment number two is is B and Rudy AB Rizal Bob Moritz
have created something called targeted proteomics that for the first time give
us real hope that we can push a human proteome project that’s absolutely
parallel to the human genome project and in fact in the middle of February we had
our first meeting trying to convince funding agencies that we should do this
and I think the meeting was intellectually successful whether
there’s any money to pay for such project as is a different topic we’ve
developed a series of clinical assays and I’ll show you a few of those that I
think are going to open up completely new possibilities for what you can do
with patients and one project I won’t talk about I call the second genome
project and that is we have access to the world’s third largest data center
and we’re going to use that to capture all human genome fully sequenced genomes
as they become available together with their attendant medical records and
phenotypic records and molecular and cellular data and we’re going to mine
those for the predictive medicine in the future and obviously you have to use use
all the latest high-level computing technologies and in the cloud to do all
the calculations and storage and that that kind of thing so family genome
sequencing we’re doing it with a company called complete genomics that for us has
done absolutely a superb job we’re doing it for about six thousand dollars a
genome this next year we will do 615 complete genomes from a whole series of
different families but let me talk about what we did in the first study which was
published in science last year we looked at a family of four the parents were
normal the kids each had two different genetic diseases and what we didn’t
realize at the time was the enormous power using the principles of Mendelian
genetics would have in in transforming our analysis and it did so in many ways
one we could use it to correct more than 70% of the DNA sequencing errors
number two we could use the family structure to identify immediately all
rare variants if two or more members of the family had the rare variant it was a
rare variant and not a sequencing error you can never make that statement if you
sequence a single individual number three we have the ability to reduce
enormous ly the data space the data search space for disease genes I’ll say
more about that in a moment and number four and this was the thing
science was far and away the most excited about we could for the first
time measure the intergenerational mutation rate in a human family and we
demonstrated it’s about 30 mutations per child and of course a really interesting
question is are all families approximately that have that rate of
mutation or will there be wide variations and we have a whole series of
other families and will soon be able to answer that question we don’t have the
answer yet what we were able to demonstrate is by reducing the disease
gene search space we were able to say unequivocally that these two genetic
diseases could only be in court encoded by one of four different genes and by
various ways we could make the proper disease gene assignment in a in a very
straightforward way so our feeling is now that for any simple Mendelian trait
family genome sequencing will give the answer straight away what we’re doing
now is we’ve sequence 65 patients from families that have Huntington’s disease
and we’re looking for a modifier gene and the modifier we’re looking for is in
those families some members get it very early in their life
others get it very late in their life so we’d like to find the gene that gets it
very late and be able to turn that on for the other family members that have
it so it isn’t a cure but you could really change the quality of the life
for many different people and the analysis is almost finished we
have data on that very soon we think also we’ll be able to take complex
genetic diseases like Alzheimer’s and if we can do the proper kind of
stratification we can convert that is if we could say there were 10 subtypes of
Alzheimer’s then each subtype would equivalent
become equivalent to looking for modifier genes and I’ll show you how
we’re going to do the stratification after I tell you about some clinical
assays a little bit later on now the really key and interesting point of
course is the amount of cost and sequencing has come down exponentially
so in 2002 it was estimated that the human genome sequence it was done cost
300 million dollars today we can get 1/4 that’s much better data or error rate
now is less than one in a million error rate so it’s spectacular and
that’s a 50 thousand fold reduction in cost so my prediction is in less than
three years we’ll be well below a thousand and I’d say in five to eight
years we’ll be in the hundreds of dollars to do a genome and the
implication of that is you’re all going to have your genomes done in the context
of your families as a part of your medical records and we’ll talk about
what that means a little bit later on so the human proteome project is one that
is be has really made four major advances to that I think really make it
possible and this obviously is the collaboration with Agilent and ABC X in
a absolutely fascinating company Chinese company Chinese American company called
foraging and the what are the four things that we’ve done number one we’ve
created a really complicated integrated software package called the transport yo
McPike line that essentially takes mass spectrometry data and evaluates its
quality initially when we created this we threw away 95% of the data that we
got from the literature and it’s improved somewhat but I’ll tell you the
quality is very country and lab dependence so you almost know what the
answers are going to be so we put together goodness
this little guy keeps popping up we put together that transfer you’ll make
pipeline with the creation of a peptide protein database the the Atlas that has
about 4 million mass spectra and mass spectrometry spectra in it and it’s used
by everybody in the proteomics world and the technology advances r21 the creation
of targeted proteomics using triple quadrupole x’ and M or M or SRM analyses
and the basic idea is that you can easily analyze between 100 and 200
proteins quantitatively in an hour if you put in radio labeled isotopes as
standards for the peptides that you analyze for each of those 1 to 200
proteins so it means we can look at biological networks you know in a
cohesive integrated way within the limits of detection for the mass
spectrometer and to show you how powerful this is Rudy Eversole and his
colleagues have done mrm assays for all 6,000 some proteins in the yeast and
they can detect them from the lowest copy protein up to the highest copied
protein absolutely beautifully and that’s exactly what we’re going to be
able to do for humans because recently Rob Moritz
has created M or M SAS SRM essays for all 20,000 human proteins not all have
been validated but they’re there to be tested for the future and that’s going
to be a key part of getting the human proteome project started now in the
future what we’d like to be able to do is identify 50 organ specific proteins
from each year 50 major organs in the blood and we’d be like to be able to
measure those 2500 proteins so we could assess
as opposed to disease in a longitudinal fashion for each patient
so what we’ve done together with Caltech is really started working on a highway
parallel platform namely micro fluidics protein chip assays and right now Jim
has a chip that one can measure 50 proteins using Eliza essays from a
fraction of a droplet of blood 300 nano leaders in a five-minute assay with a
dynamic range of close to 10 to the sixth and we have these devices in the
clinic at UCLA testing individual patients response to with melanoma and
glioblastoma to drugs and and antibody therapies by
mere fingerprint pricks and what’s cool about that is you don’t even have to go
through our bees to get fingerprint pricks so the question is how do we
scale this from 50 proteins and those are proteins whose Eliza assays have
been created by industry spending gazillions of dollars to get them and my
argument is antibodies will never work for 2,500 protein measurements so what
we’re doing together with Jim is using peptide protein capture agents trying to
develop these as as protein capture agents and the basic idea is the
following Jim is created library of 10 to the 15th d amino acid 6 MERS some of
them are branched and what you can do is take the protein you’d like to create
the capture agent against and you can use it as a screen to get low binding
monomers then what you can do is take pairs of
the monomers that are in there right three-dimensional orientation and you
can join them together to make a bivalent reagent with a chemistry called
click chemistry that’s been perfected by Barry Sharpless at Scripps and you can
actually even join at another level to get a trimer and most recently we’ve
gotten tetra Merce and with each mer you add on
you get one to two logs increase in census sensitivity and specificity and
the what is great about these reagents is they’re rock stable resistant Brodie
Asus so Jam it with the first reagent heat made put it in the trunk of his car
through the hot Pasadena summer took it out in August in September and measured
its affinity and it was unchanged so what’s great about these reagents is
there are things that we could send at the developing world quite obviously so
we’re very high on these reagents but but there are other things after MERS
are making a comeback after a long drought and so there there are other
possibilities as well let me just say that we’ve developed 16 different assays
using various combinations of genomics and proteomics and all I’ll talk just
about one assay that I’m really excited about or gene has expressed 12,000 human
proteins in embryonic human embryonic kidney cells and or gene has
demonstrated that it can take 20 to 30 of these cells and place them on a glass
slide and lyse them and you essentially have a protein reagent there and they’ve
actually created chips that have all 12,000 of these proteins and what
they’ve demonstrated is if you look in autoimmune disease you can pick up the
auto antibodies beautifully and my prediction is this is going to let us
stratify autoimmune diseases in a way we could never think about doing before
because if you take something like type 1 diabetes I mean with all the studies
that have been done there are 5 or 6 antigens that have been discovered and
I’ll guarantee of the first time we look at patients with type 1 diabetes we can
double or triple that number just because you’re doing the studies in a
global rather than a one at a time kind of way
and we could talk about any of those other essays we’re doing single-cell
analysis I’m going to tell you why that’s going to be important in just a
second and we’re going to be using induced pluripotent stem cells from
individual patients and that’s really going to be important for stratification
of disease so just to give you an idea of how single-cell analysis is really
important we actually looked at 32 cells recently from human glioblastoma cell
line we looked at 24 different transcripts in each of those cells and
then we did a multi-dimensional analysis of that those transcriptomes and we’re
able to demonstrate beautifully that those 32 cells fell into three quantized
groups with a couple of outliers that were discretely focused in quite
separate one from another now we have no idea what those discrete groups do but
that they are different distinct populations there’s no doubt about it
and with the Human Genome Atlas Project where you sequence all tumors in this
case what you’re doing is you’re destroying the signal and you’re
increasing the noise and I would argue these quantized cell populations really
may be the key to successful therapy in the future but anyway we can use
microfluidic platforms now to look at hundreds of transcripts from an
individual cell so what we’re planning to do in the future is collaborate with
a company called cellular dynamics which has two unique capacities one we can
send them five mils of blood and from the white nucleated cells they can every
time yet IPS cells and iPS cells are really important because you can
replicate them infinitely and make very large populations and that’s critical
for a lot of things we’ll want to do in the future the second thing that they’ve
learned to do is differentiate it to at least four major cell types so they can
differentiate them to myocardial Site 99% pure all of these are 99% pure and
athelia fell to parasites and most interesting they can they can
differentiate them to neurons so what we plan to do in families that have
Alzheimer’s disease is to create IPS cells from all the members of the family
to differentiate those IPS cells so we’ll have each patient’s neurons in a
test tube and then will perturb those neurons with appropriate environmental
signals because each different type of alzheimers disease will have a different
combination of disease perturbed networks and hence the environmental
perturbations will be uniquely defining for each of the different types this is
all hypothesis we haven’t done it yet but I’ll guarantee it’s going to turn
out to be true and we’re we’re moving along very quickly to do this kind of
thing so the final thing is the analytic tools are obviously really going to be
key and it’s late so I’m not going to get into a lot of that let me just say
that a third of ISB is software engineers computer scientists
theoretical physicists doing modeling engineers doing computer science and one
form so we’ve created data pipelines that capture and validate the store and
mine and analyze and integrate model global data sets and the most important
thing I can say to young students is all software should be driven by the
appropriate domain expertise problems don’t invent software in isolation from
the real problems and here are a whole series of things we’ve done in the last
couple of years that are absolutely related to specific kinds of problems
and I’m not going to get in that but rather I’ll leave you with a couple of
interesting thoughts about computational challenges in biology
you have to realize data space for organisms is infinite so you have to
figure out how to formulate the perturbations
give you relevant insights into data space that can be informative number two
high dimensionality of data enormous signal-to-noise problems and we’ve got
to figure out how to deal with that how do you separate all the biologies how do
you separate the technical kinds of things number three we need to have
global measurements of all the different data types and how do we deal with
quality assessment than their challenges how do we integrate these multi scale
types of data to create predictive models and so forth and one thing you
should really understand in biology is how much information you need should be
aimed at the nature of the problem that you want to solve I would argue
diagnostics in medicine needs infinitely less information than understanding
biological mechanisms so you want to design these these are really expensive
studies so you want to optimize your studies to get what information you need
but not to be not to be excessive about it and finally one of my favorite points
is there’s no such thing as a pathway there are only networks and I’ll tell
you I’ve seen time and time again people make major mistakes in talking about
pathways because it focuses their thinking and linear dimensions rather
than multi dimensions so as a student that’s really a key point all of these
things bring us to this thing called p4 medicine and let me just give you my
10-year view of where we’ll be with some of the dimensions of the 4ps so
predictive I predict that in 10 two or so years we’ll have our genomes done
together with some environmental information that will give us enormous
insights into the future trajectory of our health and we can advise patients in
ways we’ve never thought about advising them before with regard to wellness and
illness is going to be a key theme and p4 medicine as you’ll see in a moment
the other prediction that I would make is I think easily in ten years we’ll
have a little handheld device that can Precure some and take that droplet of
blood make 2500 measurements sended vial wireless to it to a server and analyse
the information and come back to you and your physician with your fine do it
again in six months or whatever and I think this is going to be enormous ly
important not only for detecting the onset of disease but for reinforcing the
things that optimize wellness as well what about personalized well we differ
by six million nucleotides on average one from another when I was in medical
school I could never understand why physicians would say let’s take an
average these patient populations and we’ll say everybody that falls and the
extremes is sick and I could never understand it because of genetic
variation and the fact is every patient is going to be their own control for
health and disease it has to be that’s the essence of what personalized
medicine is but you know just think about it you’ll all have millions of
data points we’ll be able to sculpt with exquisite sensitivity the dimensions of
health and wellness and predictions about the future for each of us and even
more we’ll have in America 360 million people with billions of data points and
that will give you the resources again to mine for the predictive medicine of
the future if and this is really a big if if we can get around all of the
societal constraints and IR beasts and all those things I think it’s about
50/50 that all of this mining for the p4 medicine of the future will be done in
China because Europe and the US have exactly the same impossible constraints
at this point in time unless it gets fixed it’s we’re going to lose out in a
big way like we’ve lost out on some other kinds of things the preventive
side of things is on the one hand this new approach to identifying drug targets
using drugs to put herb networks to make them behave more normally on the other
hand for the first time we’re going to really understand how to effectively
turn on innate and adaptive immunity and create vaccines that are really going to
be effective for malaria and AIDS and all the things that we’ve we’ve I
remember having an enormous argument with Tony Fauci ten years ago and my
argument was take the billions of dollars you’re going to spend on AIDS
vaccines and put a small fraction of it into systems approaches to understanding
the immune system they did that this year so they’re only ten years off but
at least they’re getting started they have wasted billions of dollars
giving it to companies that essentially did the vaccines just exactly like
Jenner did in 1796 quite frankly I mean there have been a couple of exceptions
but but the exceptions didn’t work either
so participatory deals with social aspects so how do you convince how do
you educate patients about the potential of this how do you convince the
skeptical physician how do you how do you inform the medical network that’s
going to have to carry out this what you have to realize is that p4 medicine is a
real revolution and it’s going to require changing the structure of our
healthcare delivery system in some really major ways that people are going
to feel uncomfortable with and I’ll tell you later how how we plan to approach it
in one way anyway so let me give you four societal implications number one is
this is going to force in the next ten years it will force every sector of the
healthcare industry to totally rewrite their business plans and the really
interesting question is can old companies learn to do new things and I
think in most cases they’re going to really have trouble and this is going to
be true a lot of drug companies though there’ll be enormous opportunity for
people that are at the leading edge of this to create new companies like like
integrated Diagnostics number two this is going to lead to an incredible
digitalization of Edinson and it’s a digitalization in
three different dimensions so number one how are we going to deal with these
billions of data points number two we’re actually going to be able to get
actionable information from one molecule one cell one genome one any unit of
information in biology that’s what this quantized digital kind of meaning and
number three very soon we’re going to have iPhones for health that we’ll be
recording hundreds if not thousands of our own data points and feeding back
continuous information on how our multi systems are responding in various ways
this is already starting to happen with Eric Topol at Scripps and really
interesting kinds of ways so I would argue this digitalization is going to
have an impact far greater even than the digitalization of information
technologies and and communication number three I argue that p4 medicine is
actually going to turn around the ever accelerating costs of healthcare and
they’ll come down to the point that I think we’ll be able to export it to the
developing world that is p4 medicine will be the framework for global
medicine and I think this for five reasons one Diagnostics we’ll get a
stratified disease and make ideal impedance matches against the
appropriate drugs number two we’re going to have a completely new way to make to
identify drug targets and hence a much more efficient way of creating drugs
they’re not going to cost a billion dollars anymore and we can test them
real time along the way in abort abort studies at an early because more and
more we’re going to be able to do studies with very small groups of
patients and the reason we can do that is if you take 40 patients and you have
a 98% success rate you’ve got all the power you need to convince the FDA to
make that a drug and Genentech did that for the drug called Herceptin for breast
cancer number three the benefits of wellness I think
are going to be absolutely staggering number four these technologies that are
exponentially increasing in measurement potential and decreasing and cost are
really really going to let us sculpt the dimensionality of individuals and really
interesting points and and it’s going to be done inexpensively and then number
five I think they’re going to be medical advances in stem cells neuro
degeneration aging vaccines and the like that really are going to be
transformational over the next ten years and finally I’d make an even more
positive argument this nation’s that really adopt this it’s going to be a
source of enormous wealth one it’s going to reduce the cost of health care
significantly but on the positive side in an information based economy where
people can stay well you increase enormously the productivity and
economists can do calculations and the calculations easily are in the hundreds
of billions of dollars for for that kind of thing and then finally there are it’s
going to spawn two categories of new industries a wellness industry on the
one hand and the health industry on the other end that will be totally different
from the health industry we know here so all of those are enormous opportunities
for creating wealth so the challenge of p4 medicine is twofold technical and
societal and the societal is going to be by far the bigger challenge so how do we
go about really making this happen something we’ve really thought a lot
about it is B we’re setting up key strategic partnerships to bring p4
medicine to patients and let me tell you just about two of those one is a an
agreement we made about two and a half years ago with the state of Luxembourg
that gives us 20 million a year for a five-year period to attack several of
the most fundamental problems in p4 medicine in return we do a lot of things
for them and I’d be glad to talk about those but I’m not going to talk about it
here number two we’ve created a non-profit institution called the p4
medicine Institute in partnership with Ohio State in the idea
here is that that Institute is now creating a network of health care
centers that will use these assays we’ve developed at ISB and use them in the
context of pilot projects because I’m really convinced to a skeptical world
the way you convince them is you show you show them that you can do things
that never could be done before so our first two projects are a wellness
project so we’re we’re measuring for the first time we’re developing for the
first time metrics for measuring bolus be glad to talk about those and number
two we’re going to look at heart failure which is a major problem that really has
been intractable for a long time so the essence of p4 medicine then or one
quantifying wellness in and demystifying disease and of course it’s these major
societal opportunities that we talked about the economic opportunities the
digitalization of medicine turning around the costs of medicine and finally
the creation of wealth and so forth so let me end up with four final comments
so the first is about students and how we are improperly training them this is
one of my favorite New Yorker cartoons and it says never ever think outside the
box and and the real question is how can we change our educational process so our
students will really think outside the box automatically and never as as most
of them do and of course a part of this I would argue is key cross-disciplinary
training I think this is one of the big revolutions coming and watered by all
the gene it seems to me fun I’ve heard some of you here at Iowa State or in
good shape in that regard point number two is big and small science by big
science I mean integrated systems approaches to two big problems as you
know at NIH there is enormous pressure to do away completely with big science
anytime budgets get tight all of the small
scientists and the bulk of people at NIH or small science oriented people in fact
I was just on a committee set up by General Medical Sciences to review the
four major new grants big science grants and it was really interesting to see the
composition of the committee you know they ended up arguing in
various ways but you know I hope it’s going to come out to a good end we’ll
see but the really important point is that at NIH
I would guess five or six percent of the dollars go to big science 70 or 80% of
the dollars go to small science so keep that in mind as we’re making these
arguments and I would argue there are many problems that cannot be approached
by small science at all they’re just too big
they need integrated cohesive well led efforts to attack them
and the other point I’d make is that there is enormous synergy between small
and big science because big science creates the fuel for enormous
opportunities for small science we’ve had terrific interactions with a lot of
people and small science a final point a third point is the idea that all of
these paradigm changes that we saw required new organizational structures
and that’s certainly going to be true of p4 medicine and I wonder what other ones
are going to be required and I will tell you what my plan is so I I was at the
hundred and fiftieth anniversary of MIT about two weeks ago and Eric Lander and
I gave our talks and then we’re on a panel and and someone asked me the
question well how are you how are you actually going to really bring this p4
medicine to to the world and I said well we’re going to do pilot projects in this
network of institutions that we’re setting up in the US and then what we’re
going to do is we’re going to go to Luxembourg because I know the three key
ministers and they’ve all agreed that if our pilot projects are successful we
bring p4 medicine to Luxembourg so it’s 500 thousand people it’s the highest per
capita income in the world and it’s a single-payer system so it’s an
absolutely ideal test model if it really works there then maybe this
heterogeneous mess of a system we have in the United States so Eric Lander got
up immediately said you know this is treason
why are you sending it away why don’t you do it here so I said Eric I wrote a
paper for the Obama transition team that discussed p4 medicine and health care
and I said I never even got an email back of thanks so if you have real
connections and you can get something going let me know but so anyway but
that’s what we plan to do and the final point that I would make that I’ve really
thought a lot about I don’t know how many of you read Tom Friedman’s book on
the worldís is flat but his point was essentially that there is a
globalization of education and economies and I would argue there’s a fascinating
globalization of science that’s occurring too and what this allows us to
do for the first time is to ask the question who are the best scientists in
the world to attack really hard problems and bring intense focus on solving them
because getting the best really makes a difference rather than maybe just
somebody who’s next door to you so that’s really what we’ve tried to do in
the strategic partnerships I talked about and the ones that will be in the
in the process of setting up in the future but I think it’s really also
interesting from the point of view of the US government isn’t going to fund
big science except at a very very low level and I’d say 3 to 5 million a year
is a low level particularly if you have to have and people involved in the grant
and by the time you divide it it’s a glorified r1 but if you set up these
strategic partnerships if you can get 20 million a year for five years that lets
you do things that are really nonlinear with regard to future
possibilities so lots of people have been involved I’m not going to go
through and name beyond the ones that I’ve already named but it’s been a great
partnership with with many different people on these things so thank you very
much so I’d be glad to answer questions yeah I could comment on that I mean
that’s curious we we have just begun to talk with Children’s Hospital in Seattle
about a strategic partnership that we’ll be looking at this whole problem of
maternal and fetal infant health and so forth and so what we’re setting up to do
actually is to look at several major fetus threatening conditions and to from
those placentas get the placenta specific proteins that are blood placed
in the blood so that we will be able to follow those kind of conditions from
conception all the way through to birth delivery so I I think this is an area
that almost nothing has been done and it’s an incredible area of opportunity
and I think many of the technologies that we have here would fit into that
beautifully but especially these blood protein assays that I think could let
you assess directly the placenta and indirectly the mother and the fetus and
in fact we may be able to look at maternal and fetal blood proteins that
would be interesting in this regard to so I think there are a lot of things
that could be done I think that is really an important area it’s actually
an area that the Gates Foundation is really getting interested in so we’re
hoping to get major funding to this kind of thing from them other
questions yes yeah so the question is are there patient testimonials that are
missing from this and again understand that we’re just getting started in these
things so there haven’t been any chance for patient testimonials but I’ll tell
you where there are patient testimonials that are really interesting is in doing
genome sequence analysis so companies like Navigenics and 23andme now have
quite following avid followers that that are really excited about the potential
of using genomics to be able to assess disease possibilities if you’re normal
you may be disappointed there may not be much you can see but for example 23andme
now has identified more than 6,000 family members from families that have
Parkinson’s disease now the CEO of 23andme has a husband who is susceptible
to parkinsonism so it’s a particular interest but these families are enormous
ly enthusiastic about doing whatever it takes to learn more about it I mean I
think you know the question of whether data and making your data or available I
mean as I look at what’s being done in social networks I really come to the
conclusion that if we wait for a little while that is it going to be a problem
because what they put out on facepage my goodness is is far more than then your
genome sequence and things like that but seriously what we do have to do with
these making patient records more accessible so we can do the mining is
make sure that there are laws that protect you against insurance companies
against employers actually frankly against your family to a certain extent
to that that and those are slowly being put in
place not very effectively but slowly yes well I think anytime you’re dealing
with large datasets the real problem is signal-to-noise I mean and I we still
don’t have very good ways for dealing effectively with that and I think that’s
the big one that we have to get over but sure there there is always I mean in any
complex process a biologist can get lost in the trivial details there’s
absolutely no question about it so what’s really key is to think about the
deep principles and try and keep yourself on the on the right track of
where you want to go and what you’d like to solve but yeah I know in biology is
that goals were cartoon showed I mean they’re infinite detail you could you
can get fascinated by all sorts of little things and and and and frankly
this is what happens to a lot of people who do small science they do things that
aren’t really very important because they have no broader context from which
they can make judgments about how important that is you know one of the
questions I asked on this committee was look we’re having a review of big
science now to see how well it did in these things have anyone ever reviewed
small science for example is anyone ever reviewed the fact there must be 600 labs
that are studying p53 and they’re all doing almost exactly the same things
some better but is I mean is that the best way to do things
so I mean you know it’s a very complicated multi-dimensional equation
so yes I’m sorry I have to speak up a little
bit I can’t couldn’t hear the last part well you know I think it’s less negating
than explaining the results that they find and the important question I think
isn’t saying you’re wrong because I suspect good people aren’t going to be
wrong but it’s is it important that’s the really important question we be
studying this kind of thing or not so I don’t think they’ll negate anything I
think they’ll just put it in context small science yes so will patents pose a
barrier for p4 medicine you know the interesting point is at this point in
time I don’t think so because a lot of the things that we’re thinking about are
so new and so different and let me give you an example when we first put a
patent in on organ specific proteins probably eight or ten years ago there
was nothing in the patent literature that was even remotely close to that so
what’s been very attractive is these really new ideas have a lot of space for
patents to be expanded around now where things can get complicated obviously is
with genomic information where there’s an enormous amount of data out there and
how exactly all of that is going to tie into being able to do a whole genome
sequences you know what I think is going to happen in the end is companies that
own individual genome sequences are not going to be able to sue 340 million
people because they had their sequence of the brca1 or 2 gene done so I think
just by sheer force of the magnitude of what’s going to happen those kind of
that won’t be a problem I think what we are going to see more
and more or what I call systems integrative
patents where you put together collections of information that are
uniquely defining for disease state and there’s not anything like that out there
except maybe a few multi-dimensional transcript studies genomic health and
things like that but there are much more sophisticated ways you can do a lot of
those things and I think even integrating different kinds of
information together it’s really going to be informative and interesting ways
and we’re just figuring that out right now other questions well I think the giveaway and the
paradigm change for pushing a person who is pushing the paradigm change is utter
conviction that this is the way it’s going to be and this is the right way to
do it and fortunately I’ve always had that conviction now you know I’ll tell
you just to give you an example I’d been when I went to Caltech in 1970
I told my chair I was going to spend 50% of my time doing engineering you know
developing technologies and he came in after three years and he said I’m urging
you in the strongest possible terms to give up all of this engineering and I
said no I wasn’t going to do it and and I got tenure
two months later so it didn’t have anything to do with tenure and 20 years
later I said you know why why did you say that do you remember that he did and
he said he said it because all of the senior professors in biology at Cal Tech
felt it was unseemly to have engineering and biology and he said what they
suggested I do is move you to engineering and he said at least I
didn’t make that suggestion I said yeah so you know I think you just have to be
really convinced that that what you’re doing is is right and have
and and you know you can be convinced and sometimes you’re wrong but on the
big ones it’s better to be right there’s no question about that yeah over there well my my conviction so the question is
can we use systems approaches systems engineering to approach really difficult
problems like cancer where you go through multiple events and so forth my
feeling is the key to that is single-cell analysis I’ll guarantee you
that’s going to give us enormous insights into the populations and the
transitions and the quantized groups that actually carry out some so I think
what we have to be able to do what we’d love to be able to do in the next couple
of years is to set up a microfluidic device that can do the entire
transcriptome for a thousand cells in one one-hour run then you could really
begin to take on problems like that but I think it’s it’s again getting down to
the modules of information and one of the key modules is the cell and when you
average the populations of cells you lose the signal I mean that’s just clear
again and again and again so that’s that’s the kind of way we’re thinking
about it was their question yes well there are a whole bunch of social
problems now that would be something that could be solved by going around
state bureaucracies but I’ll tell you so we’re thinking of setting up a
strategic partnership with Saudi Arabia to look at the royal family and they
have simple genetic diseases and complex genetic diseases one of the things that
they’re terrified Val is sexual incest and I mean you know these analyses will
say gee you’re really the daughter of this guy over here and not the daughter
of this guy over here and so we haven’t quite figured out how to get around that
one that’s that’s so so but but there are problems like that and in fact there
was a recent paper in Lancet that showed large-scale Chiwa studies that
absolutely identified several cases of incest and I mean that were that were
utterly clear so I mean those are the kind of things you have to be sensitive
about yeah