hi so I am a I’m using this spirit as a
scientist and these are the beautiful
creatures that my team and I study so we
study networks and the patterns between
nodes and how they am how they connect
and what I’m going to tell you is how I
got so intrigued by them in the first
place at the beginning of my scientific
career how we use them today to try to
help children with severe genetic
diseases and how I also imagine that
they will be useful in the future and
have the potential really to
revolutionize the way we do medicine and
and science in general so let me
introduce you to the the basic idea why
we think that studying networks is
something useful to begin with with this
beautiful design object from the year
1990 that is a Macintosh classic and
this is what it looks like if you take
it apart and already from this picture I
think it’s pretty clear that if you only
know the components of such a complex
system even like a computer you are not
going to be able to fully understand how
the system works and you’re also not
going to be able to repair it if
something is broken and the one crucial
component that is missing is how these
bits and pieces are actually fit
together so how are they wired and this
is only an example of a small part of
this computer and that is the the chip
that actually does the computations so
the promise of doing science of networks
is that by staring long enough at these
wiring diagrams and that the structures
that we see there we can learn something
deep about how the systems that they
represent really work now this map is
relatively ordered and tidy and the maps
that I actually study I study cells and
the human body are much more messy and
this is how they look like so this is a
still frame from the video that you that
you saw in the beginning and this is in
a way a map of a human cell so the nodes
that you see here these are proteins
little biomolecule not so little
biomolecules or molecular machines and
essentially everything we do from
digesting food to
moving our muscle to fighting infections
is on the molecular level the result of
biomolecules like this interacting with
each other so what we as network
scientists do is we try to make sense of
this pretty mess and we try to bring
order and see patterns and structure and
ultimately and see how the connection is
to the systems that they represent so
when I started at the beginning after I
studied physics and I was wondering what
to do next and this was right about the
time when the first of these maps
actually emerged so and researchers had
done the painstaking staking work of
really connect of collecting all these
intellect interactions that are
happening in a cell and around the same
time also very different maps were
emerging
so people were also starting to measure
how computers are wired together in the
internet or how neurons are wired in the
brain or how people are connecting in a
social setting and the fantastic finding
of the the pioneers of this field was
that all of these networks are on a very
deep structural level actually very very
similar as one example is this network
that we are moving from the tiny little
microscopic network of the proteins
interacting in our cell to the largest
infrastructure that humanity has built
and that is a map of the internet so the
nodes that you see here are serviced
this would be large computers
essentially and there are lines between
them if they send and receive data from
one to another so if you imagine that
you are let’s say watching a TEDx video
on YouTube what really happens is that
the video will travel from the server
where it is actually stored probably
some Google server through a number of
steps until it reaches your provider and
your phone and what this network that is
working on an entirely different scale
and the cellular network and the brain
network and many many other networks
have in common is for example that the
vast majority of the nodes in this and
other networks have only relatively few
connections most people on Facebook have
only not so many friends most servers
here would only communicate with few
other computers and also in our brain
the most cells are actually not so well
connected on the other side there are
also what we would call super hubs so on
the right side you see this bright
flowery structure or there’s there’s
another one on the left side and these
are nodes that are extremely well
connected so that spread out their
connections throughout the entire
network in a social network you could
think of Kim Kardashian or in a web site
that would be Google and this has
important consequences for how these
systems work or what also happens how
you can damage them and what we as
network scientists again try to do is to
understand how did these similarities
emerge between these entirely different
systems and how can we understand by
looking at one network also other
networks so what happens
how fast can information spread for
example in a network or how can i how
can I repair it and to illustrate you
how we try to make this connection
between network structure and what they
represent let’s have a look at the
largest social network that there is so
Facebook where the nodes are well you
and me most of you I imagine and links
in this network represent Facebook
friendship and what is already visually
immediately obvious is that you see how
the continents emerge from this map and
that is pretty straightforward it’s
there are not too many people living in
the Atlantic Ocean so there is not too
much light there not too many Facebook
users but there are also other holes so
there are other geographical holes you
can see the Sahara you can see the
Amazon basin but there are also
political holes so you can see that
there is relatively dark in China and in
Russia or economical ones let’s say
parts of Africa where in principle there
would be a lot more people living there
and then have access to an technology
like Facebook now in this example we
know all this right I mean we know that
there are continents that have different
political and economical status but the
promise is that if you look at a system
like the cell that we in fact do not
understand very well that by studying
very closely how the components of the
cell interact with each other we can
learn something also deep about how the
cell works another metaphor let’s say
that has been guiding our work over the
last couple of years is to take this
notion of a map and transported to these
networks and to illustrate that I like
to use the most iconic map there is the
one of New York City and the three key
components that are relevant to our work
would be the following the first one is
that location matters so without knowing
the numbers I imagine that well it’s
pretty straightforward to guess that the
Starbucks at Times Square sells a lot
more coffee than the Starbucks on
Roosevelt Island and actually check
there is a Starbucks on Roosevelt Island
so location is important the second
important concept is the one of
neighborhoods so if you compare let’s
say Wall Street to the beautiful area of
Williamsburg in Brooklyn these two
neighborhoods have very different
functions and are frequented by a very
different clientele I imagine and the
third one is the distance matters on
these maps so I’m without being a
political scientist I imagine that it is
no coincidence that the richest part of
Wall Street is maximally removed from
the poorest neighborhood in New York
City and up there in the Bronx so how do
we translate this to these molecular
networks where again the nodes are our
molecules our proteins and the
interactions are that they actually do
bind in our cell the first one we found
that also location matters in these
networks so if you compare these red
genes in the middle that are very highly
connected that are very central they
more often than not are also part of
essential molecular machines that are so
important that if you have a mutation in
one of these components you would
typically these would be lethal or
sometimes they can be associated with
very severe inborn diseases where is
these genes that are more at the
periphery and less strongly connected
are often redundant and mutations in
these genes would have an effect that is
also the concept of neighborhood turns
out to be important and meaningful and
useful in these molecular networks so
you could think that these genes that I
painted there and could for instance
represent all those genes and which a
variation has been found to predispose
you to develop type 2 diabetes at some
point in your life and what we found is
that these different genes that are
associated to the same disease they are
not scattered randomly all over the
place in these networks but they are
associated to very well-defined local
neighborhoods and this is incredibly
useful because once you identify there
is a certain neighborhood this helps you
to search for other genes that may also
be important for the same disease so you
can systematically screen your
neighborhoods for other processes that
may be also important and in fact the
analogy between the internet that I
mentioned before and these um cellular
networks go so far that we do use the
same algorithms to rank the genes for
their importance to a certain disease
that Google uses to rank and the
websites in their importance for a
certain search that we put in there and
lastly also the notion of distance is
meaningful in these molecular networks
in that the red and the blue disease
could correspond for instance to
diabetes and obesity so diseases that
are strongly related whereas those that
are farther away essentially have
nothing to do with the molecular reasons
for the disease of the two lower ones
alright so how does this how do we apply
that to actually concrete patients and
we work very closely with geneticists
and medical doctors and biologists in in
the department where I work and one of
the challenges that we are trying to
tackle is our rare diseases so rare
diseases are defined by being rare but
there are so many of them that actually
collectively between 2010 and 2004
suffering from one rare disease and the
majority of these rare diseases are
genetic that means that there is a
certain location in the DNA of these
patients and more often than not it’s
actually only a single location where
there is a defect or a mutation that is
responsible for the disease
the first challenge is that it is often
very very difficult to actually find
which location is responsible because
all of us we have thousands and
thousands of small variations in our DNA
if I compare mine with yours but usually
they only account for well our
individuality some people are tall some
not so much some have blue eyes well red
eyes dark eyes brown eyes and the
challenge to find the one red variation
in the genome of these patients that are
usually kids is actually quite difficult
it’s really to find the needle in the
haystack and how we use these concepts
that are introduced you before is like
this first we would map all the
variations that we find in a given
patient onto these networks so you would
check which are the genes that they
actually belong to and then we would try
to identify a neighborhood that actually
is a similar disease that we already
know so for example if we are dealing
with patients that have severe problems
in their immune system we would check
which other diseases we actually know
are causing a similar problem in the
immune system so this will give us a
hint what is the general neighborhood in
which we are more likely to find the
actual variation that is causing the
disease now what we have at this point
if we are successful in identifying this
this causal variation is essentially a
diagnosis a molecular diagnosis saying
the disease in that person is caused by
variation in gene X and that is already
incredibly useful because it helps you
understand what the molecular the the
causal chain of events might be leading
from that a letter in your DNA is
different all the way up to a problem
that manifests in in organs but what we
ultimately want to achieve is of course
we want to help these patients so we
would also like to get a therapist
Avenue out of that and the way we do
this is again using these neighborhoods
so we would systematically look around
the cause of gene genetic variation for
other diseases for which we actually
know a treatment so we would check which
of these blue circles are very very
close and have a treatment that works
and then we can start
discussion on whether that treatment
might also be applicable and to this
newly discovered disease all right just
to to end I would like to look a little
bit into the future because I showed you
many pretty pictures at least I like
them but that is not how our day-to-day
work really looks like this is and it
quite literally is so this is when I
prepared this slide I went a couple of
offices down and asked my postdoc Celine
to make a screenshot of what she was
working on right now so that is
literally how we work that is we do a
lot of math we do a lot of statistics
and we do a lot of coding and one of the
prime challenges in today’s biology and
medicine is that we are drowning in data
so for instance genome sequencing has or
is about to become so accessible and so
cheap and so easy to do that it is not a
stretch to imagine that in a couple of
years sequencing your genome will be a
standard practice if you go and see a
doctor as today is maybe measuring your
blood pressure the challenge is more
what to do with all this data or how to
interpret it what it really means and
one part of the answer is certainly this
so one part is powerful computers and
machine learning algorithms but there is
a second part and that is us so that is
the human intuition and the human
creativity and the human ability to take
complex data and complex facts and to
try to connect them relate them with one
another and to make sense out of them
and a few years ago or right now there
is another technology that is emerging
that I think will profoundly change the
way that we interact with computers and
that is a virtual reality so what we
designed to be specifically what
Sebastian one of our team members
designed is a virtual reality tool to
interact with these huge and complex
data sets so what you see here is what
he actually sees through these 3d
glasses and he can inspect individual
nodes what kind of protein is that
can ask the computer show me the disease
neighborhood of disease X and he can do
all of what I showed you before and a
whole bunch of more so my vision for how
a medical person will actually work in a
couple of years it might very well look
like this so that you go to a doctor and
you will see a specialist interacting
and making sense out of these complex
data that all of us have and to us this
is a tool that would in fact allow us to
dive into the complexity and into these
massive data and my my vision of where
this will go is really to to help no
longer be alienated by machine learning
and massive data but to foster a new and
I hope very fruitful relationship
between humans and machines in this very
interactive way and I think it’s gonna
be really cool
thank you [Applause]
scientist and these are the beautiful
creatures that my team and I study so we
study networks and the patterns between
nodes and how they am how they connect
and what I’m going to tell you is how I
got so intrigued by them in the first
place at the beginning of my scientific
career how we use them today to try to
help children with severe genetic
diseases and how I also imagine that
they will be useful in the future and
have the potential really to
revolutionize the way we do medicine and
and science in general so let me
introduce you to the the basic idea why
we think that studying networks is
something useful to begin with with this
beautiful design object from the year
1990 that is a Macintosh classic and
this is what it looks like if you take
it apart and already from this picture I
think it’s pretty clear that if you only
know the components of such a complex
system even like a computer you are not
going to be able to fully understand how
the system works and you’re also not
going to be able to repair it if
something is broken and the one crucial
component that is missing is how these
bits and pieces are actually fit
together so how are they wired and this
is only an example of a small part of
this computer and that is the the chip
that actually does the computations so
the promise of doing science of networks
is that by staring long enough at these
wiring diagrams and that the structures
that we see there we can learn something
deep about how the systems that they
represent really work now this map is
relatively ordered and tidy and the maps
that I actually study I study cells and
the human body are much more messy and
this is how they look like so this is a
still frame from the video that you that
you saw in the beginning and this is in
a way a map of a human cell so the nodes
that you see here these are proteins
little biomolecule not so little
biomolecules or molecular machines and
essentially everything we do from
digesting food to
moving our muscle to fighting infections
is on the molecular level the result of
biomolecules like this interacting with
each other so what we as network
scientists do is we try to make sense of
this pretty mess and we try to bring
order and see patterns and structure and
ultimately and see how the connection is
to the systems that they represent so
when I started at the beginning after I
studied physics and I was wondering what
to do next and this was right about the
time when the first of these maps
actually emerged so and researchers had
done the painstaking staking work of
really connect of collecting all these
intellect interactions that are
happening in a cell and around the same
time also very different maps were
emerging
so people were also starting to measure
how computers are wired together in the
internet or how neurons are wired in the
brain or how people are connecting in a
social setting and the fantastic finding
of the the pioneers of this field was
that all of these networks are on a very
deep structural level actually very very
similar as one example is this network
that we are moving from the tiny little
microscopic network of the proteins
interacting in our cell to the largest
infrastructure that humanity has built
and that is a map of the internet so the
nodes that you see here are serviced
this would be large computers
essentially and there are lines between
them if they send and receive data from
one to another so if you imagine that
you are let’s say watching a TEDx video
on YouTube what really happens is that
the video will travel from the server
where it is actually stored probably
some Google server through a number of
steps until it reaches your provider and
your phone and what this network that is
working on an entirely different scale
and the cellular network and the brain
network and many many other networks
have in common is for example that the
vast majority of the nodes in this and
other networks have only relatively few
connections most people on Facebook have
only not so many friends most servers
here would only communicate with few
other computers and also in our brain
the most cells are actually not so well
connected on the other side there are
also what we would call super hubs so on
the right side you see this bright
flowery structure or there’s there’s
another one on the left side and these
are nodes that are extremely well
connected so that spread out their
connections throughout the entire
network in a social network you could
think of Kim Kardashian or in a web site
that would be Google and this has
important consequences for how these
systems work or what also happens how
you can damage them and what we as
network scientists again try to do is to
understand how did these similarities
emerge between these entirely different
systems and how can we understand by
looking at one network also other
networks so what happens
how fast can information spread for
example in a network or how can i how
can I repair it and to illustrate you
how we try to make this connection
between network structure and what they
represent let’s have a look at the
largest social network that there is so
Facebook where the nodes are well you
and me most of you I imagine and links
in this network represent Facebook
friendship and what is already visually
immediately obvious is that you see how
the continents emerge from this map and
that is pretty straightforward it’s
there are not too many people living in
the Atlantic Ocean so there is not too
much light there not too many Facebook
users but there are also other holes so
there are other geographical holes you
can see the Sahara you can see the
Amazon basin but there are also
political holes so you can see that
there is relatively dark in China and in
Russia or economical ones let’s say
parts of Africa where in principle there
would be a lot more people living there
and then have access to an technology
like Facebook now in this example we
know all this right I mean we know that
there are continents that have different
political and economical status but the
promise is that if you look at a system
like the cell that we in fact do not
understand very well that by studying
very closely how the components of the
cell interact with each other we can
learn something also deep about how the
cell works another metaphor let’s say
that has been guiding our work over the
last couple of years is to take this
notion of a map and transported to these
networks and to illustrate that I like
to use the most iconic map there is the
one of New York City and the three key
components that are relevant to our work
would be the following the first one is
that location matters so without knowing
the numbers I imagine that well it’s
pretty straightforward to guess that the
Starbucks at Times Square sells a lot
more coffee than the Starbucks on
Roosevelt Island and actually check
there is a Starbucks on Roosevelt Island
so location is important the second
important concept is the one of
neighborhoods so if you compare let’s
say Wall Street to the beautiful area of
Williamsburg in Brooklyn these two
neighborhoods have very different
functions and are frequented by a very
different clientele I imagine and the
third one is the distance matters on
these maps so I’m without being a
political scientist I imagine that it is
no coincidence that the richest part of
Wall Street is maximally removed from
the poorest neighborhood in New York
City and up there in the Bronx so how do
we translate this to these molecular
networks where again the nodes are our
molecules our proteins and the
interactions are that they actually do
bind in our cell the first one we found
that also location matters in these
networks so if you compare these red
genes in the middle that are very highly
connected that are very central they
more often than not are also part of
essential molecular machines that are so
important that if you have a mutation in
one of these components you would
typically these would be lethal or
sometimes they can be associated with
very severe inborn diseases where is
these genes that are more at the
periphery and less strongly connected
are often redundant and mutations in
these genes would have an effect that is
also the concept of neighborhood turns
out to be important and meaningful and
useful in these molecular networks so
you could think that these genes that I
painted there and could for instance
represent all those genes and which a
variation has been found to predispose
you to develop type 2 diabetes at some
point in your life and what we found is
that these different genes that are
associated to the same disease they are
not scattered randomly all over the
place in these networks but they are
associated to very well-defined local
neighborhoods and this is incredibly
useful because once you identify there
is a certain neighborhood this helps you
to search for other genes that may also
be important for the same disease so you
can systematically screen your
neighborhoods for other processes that
may be also important and in fact the
analogy between the internet that I
mentioned before and these um cellular
networks go so far that we do use the
same algorithms to rank the genes for
their importance to a certain disease
that Google uses to rank and the
websites in their importance for a
certain search that we put in there and
lastly also the notion of distance is
meaningful in these molecular networks
in that the red and the blue disease
could correspond for instance to
diabetes and obesity so diseases that
are strongly related whereas those that
are farther away essentially have
nothing to do with the molecular reasons
for the disease of the two lower ones
alright so how does this how do we apply
that to actually concrete patients and
we work very closely with geneticists
and medical doctors and biologists in in
the department where I work and one of
the challenges that we are trying to
tackle is our rare diseases so rare
diseases are defined by being rare but
there are so many of them that actually
collectively between 2010 and 2004
suffering from one rare disease and the
majority of these rare diseases are
genetic that means that there is a
certain location in the DNA of these
patients and more often than not it’s
actually only a single location where
there is a defect or a mutation that is
responsible for the disease
the first challenge is that it is often
very very difficult to actually find
which location is responsible because
all of us we have thousands and
thousands of small variations in our DNA
if I compare mine with yours but usually
they only account for well our
individuality some people are tall some
not so much some have blue eyes well red
eyes dark eyes brown eyes and the
challenge to find the one red variation
in the genome of these patients that are
usually kids is actually quite difficult
it’s really to find the needle in the
haystack and how we use these concepts
that are introduced you before is like
this first we would map all the
variations that we find in a given
patient onto these networks so you would
check which are the genes that they
actually belong to and then we would try
to identify a neighborhood that actually
is a similar disease that we already
know so for example if we are dealing
with patients that have severe problems
in their immune system we would check
which other diseases we actually know
are causing a similar problem in the
immune system so this will give us a
hint what is the general neighborhood in
which we are more likely to find the
actual variation that is causing the
disease now what we have at this point
if we are successful in identifying this
this causal variation is essentially a
diagnosis a molecular diagnosis saying
the disease in that person is caused by
variation in gene X and that is already
incredibly useful because it helps you
understand what the molecular the the
causal chain of events might be leading
from that a letter in your DNA is
different all the way up to a problem
that manifests in in organs but what we
ultimately want to achieve is of course
we want to help these patients so we
would also like to get a therapist
Avenue out of that and the way we do
this is again using these neighborhoods
so we would systematically look around
the cause of gene genetic variation for
other diseases for which we actually
know a treatment so we would check which
of these blue circles are very very
close and have a treatment that works
and then we can start
discussion on whether that treatment
might also be applicable and to this
newly discovered disease all right just
to to end I would like to look a little
bit into the future because I showed you
many pretty pictures at least I like
them but that is not how our day-to-day
work really looks like this is and it
quite literally is so this is when I
prepared this slide I went a couple of
offices down and asked my postdoc Celine
to make a screenshot of what she was
working on right now so that is
literally how we work that is we do a
lot of math we do a lot of statistics
and we do a lot of coding and one of the
prime challenges in today’s biology and
medicine is that we are drowning in data
so for instance genome sequencing has or
is about to become so accessible and so
cheap and so easy to do that it is not a
stretch to imagine that in a couple of
years sequencing your genome will be a
standard practice if you go and see a
doctor as today is maybe measuring your
blood pressure the challenge is more
what to do with all this data or how to
interpret it what it really means and
one part of the answer is certainly this
so one part is powerful computers and
machine learning algorithms but there is
a second part and that is us so that is
the human intuition and the human
creativity and the human ability to take
complex data and complex facts and to
try to connect them relate them with one
another and to make sense out of them
and a few years ago or right now there
is another technology that is emerging
that I think will profoundly change the
way that we interact with computers and
that is a virtual reality so what we
designed to be specifically what
Sebastian one of our team members
designed is a virtual reality tool to
interact with these huge and complex
data sets so what you see here is what
he actually sees through these 3d
glasses and he can inspect individual
nodes what kind of protein is that
can ask the computer show me the disease
neighborhood of disease X and he can do
all of what I showed you before and a
whole bunch of more so my vision for how
a medical person will actually work in a
couple of years it might very well look
like this so that you go to a doctor and
you will see a specialist interacting
and making sense out of these complex
data that all of us have and to us this
is a tool that would in fact allow us to
dive into the complexity and into these
massive data and my my vision of where
this will go is really to to help no
longer be alienated by machine learning
and massive data but to foster a new and
I hope very fruitful relationship
between humans and machines in this very
interactive way and I think it’s gonna
be really cool
thank you [Applause]
Please follow and like us: