[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.00,0:00:18.22,Default,,0000,0000,0000,,{\i1}35C3 preroll music{\i0}
Dialogue: 0,0:00:18.22,0:00:23.83,Default,,0000,0000,0000,,Herald Angel: What happens if you mix\NShannon's information theory and
Dialogue: 0,0:00:23.83,0:00:31.03,Default,,0000,0000,0000,,biological systems? A dish better served\Nhot. Please welcome our computational
Dialogue: 0,0:00:31.03,0:00:37.68,Default,,0000,0000,0000,,systems biology chef, who will guide you\Nthrough investigating the information flow
Dialogue: 0,0:00:37.68,0:00:43.30,Default,,0000,0000,0000,,in living systems. Please welcome with a\Nvery warm round of applause Jürgen Pahle.
Dialogue: 0,0:00:43.30,0:00:51.53,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:00:51.53,0:00:56.67,Default,,0000,0000,0000,,Jürgen Pahle: Thanks a lot and thanks for\Nhaving me. It's great that so many of you
Dialogue: 0,0:00:56.67,0:01:01.46,Default,,0000,0000,0000,,are interested in that topic, which is not\Nabout technical systems but actually
Dialogue: 0,0:01:01.46,0:01:07.23,Default,,0000,0000,0000,,biological cells. So, I am leading a\Ngroup in Heidelberg at the university
Dialogue: 0,0:01:07.23,0:01:16.79,Default,,0000,0000,0000,,there and we are mostly interested in how\Ninformation is processed, sensed, stored,
Dialogue: 0,0:01:16.79,0:01:25.19,Default,,0000,0000,0000,,communicated between biological cells. And\Nwe are interested in that because it's not
Dialogue: 0,0:01:25.19,0:01:30.62,Default,,0000,0000,0000,,obvious that they actually manage to do\Nthat in a reliable fashion. They don't
Dialogue: 0,0:01:30.62,0:01:35.30,Default,,0000,0000,0000,,have transistors. They only can use their\Nmolecules mostly proteins, big molecules
Dialogue: 0,0:01:35.30,0:01:44.67,Default,,0000,0000,0000,,that are little engines or little motors\Nin the cell that allow them to fulfill
Dialogue: 0,0:01:44.67,0:01:51.88,Default,,0000,0000,0000,,their biological functions. If information\Nprocessing fails in cells, you get diseases
Dialogue: 0,0:01:51.88,0:02:00.35,Default,,0000,0000,0000,,like epilepsy, cancer and of course\Nothers. Now, cellular signaling pathways
Dialogue: 0,0:02:00.35,0:02:07.20,Default,,0000,0000,0000,,have been studied in some detail - mostly\Nsingle pathways. More and more also
Dialogue: 0,0:02:07.20,0:02:14.58,Default,,0000,0000,0000,,networks of pathways but surprisingly\Nlittle conceptual work has been
Dialogue: 0,0:02:14.58,0:02:19.19,Default,,0000,0000,0000,,done on them. So we know the molecules\Nthat are involved, we know how they
Dialogue: 0,0:02:19.19,0:02:27.98,Default,,0000,0000,0000,,react, how they combine to build these\Npathways. But we don't know how, actually,
Dialogue: 0,0:02:27.98,0:02:35.69,Default,,0000,0000,0000,,information is transferred or communicated\Nacross these pathways and we intend to
Dialogue: 0,0:02:35.69,0:02:42.78,Default,,0000,0000,0000,,fill that gap in our group. And, of\Ncourse, first we have to we have to model
Dialogue: 0,0:02:42.78,0:02:51.73,Default,,0000,0000,0000,,these networks, we have to model these \Nbiochemical pathways. And this is how we
Dialogue: 0,0:02:51.73,0:02:57.48,Default,,0000,0000,0000,,proceed. So you have a you have a cell -\Nyou can't see that here - but on the upper
Dialogue: 0,0:02:57.48,0:03:02.20,Default,,0000,0000,0000,,left corner you have that scheme of a cell\Nwith all the different components. You
Dialogue: 0,0:03:02.20,0:03:08.91,Default,,0000,0000,0000,,have volumes in the cell where\Nchemical reactions happen. So chemical
Dialogue: 0,0:03:08.91,0:03:15.12,Default,,0000,0000,0000,,reactions take biochemical species: ions,\Nproteins, what have you, and they convert
Dialogue: 0,0:03:15.12,0:03:20.40,Default,,0000,0000,0000,,them into other chemical species, and\Nthese reactions happen in the different
Dialogue: 0,0:03:20.40,0:03:26.85,Default,,0000,0000,0000,,compartments. Now it's very important to\Nassign speeds or velocities to these
Dialogue: 0,0:03:26.85,0:03:33.34,Default,,0000,0000,0000,,reactions because these speeds determine\Nhow fast the reactions happen and how the
Dialogue: 0,0:03:33.34,0:03:39.04,Default,,0000,0000,0000,,dynamic behavior then results. And once\Nyou have done that, you can translate all
Dialogue: 0,0:03:39.04,0:03:44.93,Default,,0000,0000,0000,,of that into a mathematical model like the\None shown here on the right. This is an
Dialogue: 0,0:03:44.93,0:03:49.31,Default,,0000,0000,0000,,ordinary differential equation system, I\Ndon't want to go into detail. I only have
Dialogue: 0,0:03:49.31,0:03:55.51,Default,,0000,0000,0000,,like two or three formulas that might\Nbe interesting for you. So this is just
Dialogue: 0,0:03:55.51,0:04:00.87,Default,,0000,0000,0000,,any mathematical model you have\Nof these systems and then you can start
Dialogue: 0,0:04:00.87,0:04:05.56,Default,,0000,0000,0000,,analyzing them. You can ask questions\Nlike: "How does the system change over
Dialogue: 0,0:04:05.56,0:04:10.49,Default,,0000,0000,0000,,time?" That's simulation. "Which parts\Ninfluence the behavior most?" "What other
Dialogue: 0,0:04:10.49,0:04:16.81,Default,,0000,0000,0000,,stable states? Do you have oscillations,\Ndo you have a steady state?" and so on.
Dialogue: 0,0:04:16.81,0:04:22.31,Default,,0000,0000,0000,,Now, you don't have to do that by hand,\Nbecause we are actually also developing
Dialogue: 0,0:04:22.31,0:04:27.71,Default,,0000,0000,0000,,software - that's just another thing. I\Nguess you know that all models are wrong.
Dialogue: 0,0:04:27.71,0:04:33.81,Default,,0000,0000,0000,,We try to build useful ones. So I said you\Ndon't have to do this by hand because we
Dialogue: 0,0:04:33.81,0:04:40.62,Default,,0000,0000,0000,,are also into method development and we\Nare building scientific software. One of
Dialogue: 0,0:04:40.62,0:04:44.62,Default,,0000,0000,0000,,the softwares we build is called COPASI:\NCOmplex PAthway SImulator. It's free and
Dialogue: 0,0:04:44.62,0:04:50.26,Default,,0000,0000,0000,,open source, you can all go to that\Nwebsite, download it, play around with it
Dialogue: 0,0:04:50.26,0:04:58.53,Default,,0000,0000,0000,,if you want. Because we also use more\Ndemanding computations which we send to
Dialogue: 0,0:04:58.53,0:05:03.12,Default,,0000,0000,0000,,compute clusters, we also developed a\Nscripting interface for COPASI, which is
Dialogue: 0,0:05:03.12,0:05:09.64,Default,,0000,0000,0000,,called CoRC, the COPASI R connector. And\Nthis allows you to use the COPASI backend
Dialogue: 0,0:05:09.64,0:05:15.56,Default,,0000,0000,0000,,with all the different tools that are in\NCOPASI from your R programming environment
Dialogue: 0,0:05:15.56,0:05:21.14,Default,,0000,0000,0000,,and then you can build workflows and send\Nthem to compute cluster. We think it's
Dialogue: 0,0:05:21.14,0:05:28.28,Default,,0000,0000,0000,,easy to use. If you play around with it\Nand you get stuck, then just let me know.
Dialogue: 0,0:05:28.28,0:05:31.42,Default,,0000,0000,0000,,So this is software you can use, you can play\Naround with. And where do we get the
Dialogue: 0,0:05:31.42,0:05:37.34,Default,,0000,0000,0000,,models? Well, there is a model database\Nthat is called Biomodels.net, also free to
Dialogue: 0,0:05:37.34,0:05:41.77,Default,,0000,0000,0000,,use, you can go there and download models.\NAt the moment they have almost 800
Dialogue: 0,0:05:41.77,0:05:47.53,Default,,0000,0000,0000,,different manually curated models, and\Nalmost ten times of that that are built
Dialogue: 0,0:05:47.53,0:05:52.76,Default,,0000,0000,0000,,automatically. You can just download them\Nin the so-called SBML format, which is the
Dialogue: 0,0:05:52.76,0:05:59.56,Default,,0000,0000,0000,,Systems Biology Markup Language, then\Nimport it into COPASI or other software and
Dialogue: 0,0:05:59.56,0:06:01.86,Default,,0000,0000,0000,,play around with them.
Dialogue: 0,0:06:01.86,0:06:08.29,Default,,0000,0000,0000,,OK, so coming back to biology, \None of our favorite systems is
Dialogue: 0,0:06:08.29,0:06:13.65,Default,,0000,0000,0000,,calcium signaling. Calcium signaling\Nworks roughly like this: You have these
Dialogue: 0,0:06:13.65,0:06:20.86,Default,,0000,0000,0000,,little - I mean the oval thing is a cell -\Nthen you have these red cones, that are
Dialogue: 0,0:06:20.86,0:06:26.49,Default,,0000,0000,0000,,hormones, and other substances that you\Nhave in your bloodstream or somewhere
Dialogue: 0,0:06:26.49,0:06:32.12,Default,,0000,0000,0000,,outside the cell. They bind to these black\Nthings, which are receptors on the cell
Dialogue: 0,0:06:32.12,0:06:37.26,Default,,0000,0000,0000,,membrane. And then a cascade of\Nprocesses happens that in the end leads to
Dialogue: 0,0:06:37.26,0:06:43.96,Default,,0000,0000,0000,,an in-stream of calcium ions, these blue\Nballs, from the ER - which is not
Dialogue: 0,0:06:43.96,0:06:47.62,Default,,0000,0000,0000,,emergency room, but endoplasmatic\Nreticulum, which is one of the
Dialogue: 0,0:06:47.62,0:06:52.98,Default,,0000,0000,0000,,compartments in the cell - into the the\Nmain compartment, the cytosol of the cell.
Dialogue: 0,0:06:52.98,0:06:58.46,Default,,0000,0000,0000,,And also calcium streams into the cell\Nfrom outside the cell. And this leads to a
Dialogue: 0,0:06:58.46,0:07:04.59,Default,,0000,0000,0000,,sharp increase of the concentration of\Ncalcium, until it's pumped out again. There
Dialogue: 0,0:07:04.59,0:07:10.01,Default,,0000,0000,0000,,are pumps that take calcium ions and\Nremove them from the cytosol, and pump
Dialogue: 0,0:07:10.01,0:07:16.01,Default,,0000,0000,0000,,them out of the cell and back into the ER.\NThis is very important because calcium is
Dialogue: 0,0:07:16.01,0:07:21.30,Default,,0000,0000,0000,,a very versatile second messenger. That's\Nwhat they call it. It regulates
Dialogue: 0,0:07:21.30,0:07:25.86,Default,,0000,0000,0000,,a number of very important cellular\Nprocesses. If you move your muscles, your
Dialogue: 0,0:07:25.86,0:07:31.21,Default,,0000,0000,0000,,muscle contraction is regulated by\Ncalcium, learning, secretion of
Dialogue: 0,0:07:31.21,0:07:37.17,Default,,0000,0000,0000,,neurotransmitters, transmitters in your\Nbrain, fertilization. A lot of different
Dialogue: 0,0:07:37.17,0:07:45.78,Default,,0000,0000,0000,,things are regulated by calcium and, if you\Nsimulate the dynamic processes, you get
Dialogue: 0,0:07:45.78,0:07:51.10,Default,,0000,0000,0000,,behavior like that. Here you can see it\Noscillates, it shows these regular spikes.
Dialogue: 0,0:07:51.10,0:07:59.23,Default,,0000,0000,0000,,So this is the calcium concentration over\Ntime. Now, if you actually measure this in
Dialogue: 0,0:07:59.23,0:08:06.38,Default,,0000,0000,0000,,real cells, and this is data measured by\Ncollaboration partners of mine in England,
Dialogue: 0,0:08:06.38,0:08:12.04,Default,,0000,0000,0000,,you see it's not that smooth. You\Nget these differences in amplitude of the
Dialogue: 0,0:08:12.04,0:08:17.61,Default,,0000,0000,0000,,peaks, you get secondary spikes, you get\Nfluctuations around the basal level, and
Dialogue: 0,0:08:17.61,0:08:22.85,Default,,0000,0000,0000,,this is because you have random\Nfluctuations in your system. Intrinsic
Dialogue: 0,0:08:22.85,0:08:27.64,Default,,0000,0000,0000,,random fluctuations that are just due to\Nrandom fluctuations in the timings of
Dialogue: 0,0:08:27.64,0:08:33.44,Default,,0000,0000,0000,,single reactive events. Single reactions,\Nbiochemical reactions that happen. And in
Dialogue: 0,0:08:33.44,0:08:37.42,Default,,0000,0000,0000,,in order to capture this behavior,\Nbecause this behavior is
Dialogue: 0,0:08:37.42,0:08:42.18,Default,,0000,0000,0000,,important, that can hamper reliable\Ninformation transfer, we have to resort to
Dialogue: 0,0:08:42.18,0:08:47.76,Default,,0000,0000,0000,,special simulation algorithms, for example\Nthe so-called Gillespie algorithm. And if
Dialogue: 0,0:08:47.76,0:08:51.64,Default,,0000,0000,0000,,you do that and apply it to the calcium\Nsystem, you can see you can actually
Dialogue: 0,0:08:51.64,0:08:57.00,Default,,0000,0000,0000,,capture these secondary peaks and all the\Ndifferent other fluctuations you have in
Dialogue: 0,0:08:57.00,0:09:03.56,Default,,0000,0000,0000,,there. Now, this is just a Monte Carlo\Nsimulation. I say "just". It's really time
Dialogue: 0,0:09:03.56,0:09:07.38,Default,,0000,0000,0000,,consuming and demanding, because you have\Nto calculate each and every single
Dialogue: 0,0:09:07.38,0:09:12.13,Default,,0000,0000,0000,,reactive event in the cell. And that takes\Na lot of time. That's why we do that on a
Dialogue: 0,0:09:12.13,0:09:16.76,Default,,0000,0000,0000,,compute cluster. I told you already, that\Ncalcium is a very versatile second
Dialogue: 0,0:09:16.76,0:09:21.87,Default,,0000,0000,0000,,messenger. So you have very many different\Ntriggers of a calcium response in the
Dialogue: 0,0:09:21.87,0:09:27.58,Default,,0000,0000,0000,,cell, things that lead to a certain calcium\Ndynamics. And on the other hand,
Dialogue: 0,0:09:27.58,0:09:33.24,Default,,0000,0000,0000,,downstream, calcium regulates many\Ndifferent things. And so you have this
Dialogue: 0,0:09:33.24,0:09:37.94,Default,,0000,0000,0000,,hourglass or bow tie structure, and that's\Nwhy people have speculated about the
Dialogue: 0,0:09:37.94,0:09:46.44,Default,,0000,0000,0000,,calcium code: How can it be, that the\Nproteins - I should go back - that actually do
Dialogue: 0,0:09:46.44,0:09:53.55,Default,,0000,0000,0000,,all these cellular functions - [Softly]\Nsorry - these green cylinders that bind
Dialogue: 0,0:09:53.55,0:09:58.86,Default,,0000,0000,0000,,calcium and are then activated or\Ninhibited by it, how can it be that they
Dialogue: 0,0:09:58.86,0:10:06.91,Default,,0000,0000,0000,,know, which stimulus or which hormone is\Noutside of the cell? They don't see them,
Dialogue: 0,0:10:06.91,0:10:12.79,Default,,0000,0000,0000,,because there is a cell membrane around\Nthe cell, around the cytosol. So people
Dialogue: 0,0:10:12.79,0:10:19.93,Default,,0000,0000,0000,,have speculated: Is there information\Nencoded in the specific calcium waveform?
Dialogue: 0,0:10:19.93,0:10:28.04,Default,,0000,0000,0000,,Is there calcium code? And how can it be\Nthat the proteins actually decode that code?
Dialogue: 0,0:10:28.04,0:10:35.34,Default,,0000,0000,0000,,It's fairly established, that\Ncalcium shows amplitude modulation. So the
Dialogue: 0,0:10:35.34,0:10:40.75,Default,,0000,0000,0000,,higher the amplitude of calcium, the more\Nactive get some proteins. It also shows
Dialogue: 0,0:10:40.75,0:10:45.87,Default,,0000,0000,0000,,frequency modulation, meaning the higher\Nthe frequency of the calcium oscillations,
Dialogue: 0,0:10:45.87,0:10:50.26,Default,,0000,0000,0000,,the more active get some proteins. But, maybe, \Nthere are other information carrying
Dialogue: 0,0:10:50.26,0:10:57.47,Default,,0000,0000,0000,,features in the waveform, like duration,\Nwaveform timing and so on. Now a doctoral
Dialogue: 0,0:10:57.47,0:11:02.06,Default,,0000,0000,0000,,student in my group, Arne Schoch, has\Nlooked into frequency modulation and he
Dialogue: 0,0:11:02.06,0:11:06.66,Default,,0000,0000,0000,,actually showed that there are proteins,\Nin that case NFAT, which is the nuclear
Dialogue: 0,0:11:06.66,0:11:12.50,Default,,0000,0000,0000,,factor of activated T-cells, which are\Nimportant in your immune system. They only
Dialogue: 0,0:11:12.50,0:11:17.79,Default,,0000,0000,0000,,react to calcium oscillations of a certain\Nfrequency. So they they get activated in a
Dialogue: 0,0:11:17.79,0:11:25.02,Default,,0000,0000,0000,,very narrow frequency band, and that's why\Nwe call it band-pass activation.
Dialogue: 0,0:11:25.02,0:11:32.59,Default,,0000,0000,0000,,Okay, so I guess you all know signaling speeds of\Ntechnical systems, they are fairly fast by
Dialogue: 0,0:11:32.59,0:11:37.17,Default,,0000,0000,0000,,now. One of our results, because we\Nquantify actually information transfer, is
Dialogue: 0,0:11:37.17,0:11:42.21,Default,,0000,0000,0000,,that calcium signalling operates at\Nroughly point four bits per second. If you
Dialogue: 0,0:11:42.21,0:11:47.06,Default,,0000,0000,0000,,compare that to technical systems, that\Nseems very low, but maybe that's enough
Dialogue: 0,0:11:47.06,0:11:52.78,Default,,0000,0000,0000,,for all the functions that a cell has to\Nfulfill. So how did we arrive at this result?
Dialogue: 0,0:11:52.78,0:11:58.67,Default,,0000,0000,0000,,Well, we used information theory,\Nclassical information theory, pioneered by
Dialogue: 0,0:11:58.67,0:12:05.60,Default,,0000,0000,0000,,people like Claude Shannon in the 40s,\Nalso by Hartley, Tuckey and a few other people.
Dialogue: 0,0:12:05.60,0:12:09.31,Default,,0000,0000,0000,,So, they looked at technical\Nsystems, and they have this prototypical
Dialogue: 0,0:12:09.31,0:12:14.37,Default,,0000,0000,0000,,communication system, where there is an\Ninformation source on the left side,
Dialogue: 0,0:12:14.37,0:12:19.23,Default,,0000,0000,0000,,then this information is somehow encoded.\NIt's transmitted over a noisy channel
Dialogue: 0,0:12:19.23,0:12:24.58,Default,,0000,0000,0000,,where the message is scrambled. Then it's\Nreceived by a receiver, decoded, and then
Dialogue: 0,0:12:24.58,0:12:30.08,Default,,0000,0000,0000,,hopefully you get the same message at the\Ndestination, that was chosen at the
Dialogue: 0,0:12:30.08,0:12:37.54,Default,,0000,0000,0000,,information source. And in our case we\Nlook at calcium as an information source
Dialogue: 0,0:12:37.54,0:12:45.70,Default,,0000,0000,0000,,and we study how much information is\Nactually transferred to downstream proteins.
Dialogue: 0,0:12:45.70,0:12:53.25,Default,,0000,0000,0000,,How do you do that? Well, information \Ntheory 101. Information theory primer.
Dialogue: 0,0:12:53.25,0:12:59.06,Default,,0000,0000,0000,,In statistical information theory\Nof the Shannon type, you look at random
Dialogue: 0,0:12:59.06,0:13:04.17,Default,,0000,0000,0000,,variables. You look at events that have a\Ncertain probability of happening. So let's
Dialogue: 0,0:13:04.17,0:13:12.77,Default,,0000,0000,0000,,say you have an event that has a\Nprobability of happening, and then Shannon
Dialogue: 0,0:13:12.77,0:13:19.80,Default,,0000,0000,0000,,said that the information content of this\Nevent should be the negative logarithm -
Dialogue: 0,0:13:19.80,0:13:24.80,Default,,0000,0000,0000,,which is shown here, the curve on the\Nright hand side - should be the
Dialogue: 0,0:13:24.80,0:13:29.80,Default,,0000,0000,0000,,negative logarithm of the probability,\Nmeaning that if an event happens all the
Dialogue: 0,0:13:29.80,0:13:34.70,Default,,0000,0000,0000,,time - and I will show you an example\Nlater - there is no information content.
Dialogue: 0,0:13:34.70,0:13:39.30,Default,,0000,0000,0000,,The information content is zero. There is\Nno surprise, if that event happens, because
Dialogue: 0,0:13:39.30,0:13:45.25,Default,,0000,0000,0000,,it happens all the time, it's like there's\Na sunny day somewhere in the desert.
Dialogue: 0,0:13:45.25,0:13:51.94,Default,,0000,0000,0000,,However, if you go to lower probabilities,\Nthen the surprisal becomes bigger and the
Dialogue: 0,0:13:51.94,0:13:58.69,Default,,0000,0000,0000,,information content rises. Now, in a system\Nyou have several events that are possible.
Dialogue: 0,0:13:58.69,0:14:02.47,Default,,0000,0000,0000,,And if you take the average uncertainty of\Nall possible events you get something that
Dialogue: 0,0:14:02.47,0:14:08.08,Default,,0000,0000,0000,,Shannon called entropy. This is still not\Ninformation, because information is a
Dialogue: 0,0:14:08.08,0:14:12.46,Default,,0000,0000,0000,,difference in entropy. So you have to\Ncalculate the entropy of a system, and
Dialogue: 0,0:14:12.46,0:14:17.97,Default,,0000,0000,0000,,then you calculate the entropy that is\Nremaining after an observation, say. And
Dialogue: 0,0:14:17.97,0:14:24.01,Default,,0000,0000,0000,,this difference is the information gained\Nby the observation. Now, coming to a
Dialogue: 0,0:14:24.01,0:14:28.45,Default,,0000,0000,0000,,simple example, let's say we have a very\Nsimple weather system where you can only
Dialogue: 0,0:14:28.45,0:14:33.97,Default,,0000,0000,0000,,have rainy and sunny days. And let's say\Nthey are equally likely. So you have a
Dialogue: 0,0:14:33.97,0:14:46.23,Default,,0000,0000,0000,,probability of 50%, the average of the\Nnegative logarithm is 1. So, when you
Dialogue: 0,0:14:46.23,0:14:51.58,Default,,0000,0000,0000,,observe the weather in the system, you gain\None bit per day. You can also think of
Dialogue: 0,0:14:51.58,0:14:57.47,Default,,0000,0000,0000,,bits as the information you need, or a\Ncell needs, to answer or decide on one yes
Dialogue: 0,0:14:57.47,0:15:07.64,Default,,0000,0000,0000,,or no question. Now, if it's always sunny\Nand no rain, then you get zero information
Dialogue: 0,0:15:07.64,0:15:13.18,Default,,0000,0000,0000,,content or uncertainty. The average is\Nzero. So you don't get any information if
Dialogue: 0,0:15:13.18,0:15:20.43,Default,,0000,0000,0000,,you observe the weather in the desert, say.\N80/20: You get a certain bit number per
Dialogue: 0,0:15:20.43,0:15:29.93,Default,,0000,0000,0000,,day, in that case .64 per day, and you\Ncan do that for Leipzig. In that case,
Dialogue: 0,0:15:29.93,0:15:34.43,Default,,0000,0000,0000,,Leipzig has ninety nine rainy days per\Nyear, according to the Deutsche
Dialogue: 0,0:15:34.43,0:15:39.76,Default,,0000,0000,0000,,Wetterdienst. This gives you an\Ninformation of .84 bit per day. You can do
Dialogue: 0,0:15:39.76,0:15:44.31,Default,,0000,0000,0000,,it in a general way. So let's say you have\None event with a probability of p and
Dialogue: 0,0:15:44.31,0:15:49.58,Default,,0000,0000,0000,,another event with a probability of 1\Nminus p and then you get this curve, which
Dialogue: 0,0:15:49.58,0:15:56.57,Default,,0000,0000,0000,,shows you that the information content is\Nactually maximal if you have maximal
Dialogue: 0,0:15:56.57,0:16:02.15,Default,,0000,0000,0000,,uncertainty, if you have equally likely\Nevents. If you have more possible events -
Dialogue: 0,0:16:02.15,0:16:07.74,Default,,0000,0000,0000,,in that case four different ones: sunny,\Ncloudy, rainy, and thunderstorm - you get
Dialogue: 0,0:16:07.74,0:16:12.06,Default,,0000,0000,0000,,two bit and this is because of the\Nlogarithm. So if you have double the
Dialogue: 0,0:16:12.06,0:16:18.80,Default,,0000,0000,0000,,amount of events and they are equally likely\Nyou get one bit more. Hope I didn't lose
Dialogue: 0,0:16:18.80,0:16:25.72,Default,,0000,0000,0000,,anyone? Now we are always looking at\Nprocesses, dynamic things, things that
Dialogue: 0,0:16:25.72,0:16:30.35,Default,,0000,0000,0000,,change over time, and if we look at\Nprocesses we have to look at transition
Dialogue: 0,0:16:30.35,0:16:34.69,Default,,0000,0000,0000,,probabilities. So we have to change\Nprobabilities to transition probabilities
Dialogue: 0,0:16:34.69,0:16:42.63,Default,,0000,0000,0000,,and you can summarize them in a matrix. So\Nlet's say, if we have a sunny day today,
Dialogue: 0,0:16:42.63,0:16:47.73,Default,,0000,0000,0000,,it's more likely that it's also sunny\Ntomorrow and less likely that it's
Dialogue: 0,0:16:47.73,0:16:52.36,Default,,0000,0000,0000,,raining, maybe only 25 percent. And, if\Nit's rainy today, you can't tell, it's
Dialogue: 0,0:16:52.36,0:17:01.75,Default,,0000,0000,0000,,equally likely. These processes are also\Ncalled Markov process. Markov was a
Dialogue: 0,0:17:01.75,0:17:07.61,Default,,0000,0000,0000,,Russian mathematician and you have them\Neverywhere. These Markovian processes are
Dialogue: 0,0:17:07.61,0:17:13.13,Default,,0000,0000,0000,,used in your cell phones, in your hard\Ndrives, they're used for error correction,
Dialogue: 0,0:17:13.13,0:17:20.05,Default,,0000,0000,0000,,the page rank algorithm of Google is one\Nbig Markov process. So, you're using them
Dialogue: 0,0:17:20.05,0:17:29.26,Default,,0000,0000,0000,,all the time, nothing technological would\Nwork nowadays without them. Because we
Dialogue: 0,0:17:29.26,0:17:36.77,Default,,0000,0000,0000,,have knowledge about today's weather, the\Nuncertainty about tomorrow's weather decreases.
Dialogue: 0,0:17:36.77,0:17:45.91,Default,,0000,0000,0000,,So now we have an entropy rate,\Ninstead of an entropy. The difference is,
Dialogue: 0,0:17:45.91,0:17:51.03,Default,,0000,0000,0000,,again, the information you gain by today's\Nweather. You can do the maths in our
Dialogue: 0,0:17:51.03,0:17:58.72,Default,,0000,0000,0000,,example. The entropy would be .92 bit per\Nday and the entropy rate, given that you
Dialogue: 0,0:17:58.72,0:18:05.58,Default,,0000,0000,0000,,know today's weather, is less. It's .87\Nbit per day. Now, to complicate things a
Dialogue: 0,0:18:05.58,0:18:11.49,Default,,0000,0000,0000,,bit more, maybe, we also look at a second\Nprocess in that case air pressure and you
Dialogue: 0,0:18:11.49,0:18:17.03,Default,,0000,0000,0000,,can measure air pressure with these little\Ndevices, the barometers and maybe, if it's
Dialogue: 0,0:18:17.03,0:18:22.10,Default,,0000,0000,0000,,sunny today and the air pressure is high,\Nin 90 percent you get a sunny day
Dialogue: 0,0:18:22.10,0:18:26.44,Default,,0000,0000,0000,,tomorrow. Normally in 10 percent of the\Ncases you get a rainy day and so on you
Dialogue: 0,0:18:26.44,0:18:32.39,Default,,0000,0000,0000,,can go through the table. In our case, I\Nlooked it up yesterday. We had a high air
Dialogue: 0,0:18:32.39,0:18:39.31,Default,,0000,0000,0000,,pressure and it was raining. So in our\Nlittle model system it would mean, that
Dialogue: 0,0:18:39.31,0:18:47.50,Default,,0000,0000,0000,,it's sunny today. Now, I told you\Ninformation is a decrease in uncertainty.
Dialogue: 0,0:18:47.50,0:18:51.91,Default,,0000,0000,0000,,How much information do we get by the\Nbarometer, by knowing the air pressure?
Dialogue: 0,0:18:51.91,0:18:56.41,Default,,0000,0000,0000,,This is the difference in uncertainty\Nwithout barometer and with the barometer
Dialogue: 0,0:18:56.41,0:19:00.97,Default,,0000,0000,0000,,and in our case we have to assume that the\Nprobability of high and low air pressure
Dialogue: 0,0:19:00.97,0:19:08.16,Default,,0000,0000,0000,,is the same. And we get .39 bit per day,\Nthat we gain by looking at the air
Dialogue: 0,0:19:08.16,0:19:12.64,Default,,0000,0000,0000,,pressure. Now, what does that have to do\Nwith biological systems? Well we have two
Dialogue: 0,0:19:12.64,0:19:17.17,Default,,0000,0000,0000,,processes. We have a calcium process that\Nshows some dynamics and we have the
Dialogue: 0,0:19:17.17,0:19:22.61,Default,,0000,0000,0000,,process of an activated protein that does\Nsomething in the cell. So we can look at
Dialogue: 0,0:19:22.61,0:19:28.28,Default,,0000,0000,0000,,both of these and then calculate how much\Ninformation is actually transferred from
Dialogue: 0,0:19:28.28,0:19:32.74,Default,,0000,0000,0000,,calcium to the protein. How much\Nuncertainty do we lose about the
Dialogue: 0,0:19:32.74,0:19:36.93,Default,,0000,0000,0000,,protein dynamics, if we know the calcium\Ndynamics? This is mathematically exactly
Dialogue: 0,0:19:36.93,0:19:43.01,Default,,0000,0000,0000,,what we are doing and this is called\Ntransfer entropy. It's an information-
Dialogue: 0,0:19:43.01,0:19:49.36,Default,,0000,0000,0000,,theoretic measure developed by Thomas\NSchreiber in 2000. There are some
Dialogue: 0,0:19:49.36,0:19:55.60,Default,,0000,0000,0000,,practical complications, that we are\Nworking on, and this is what we are using
Dialogue: 0,0:19:55.60,0:20:00.90,Default,,0000,0000,0000,,actually for the calculations. So in our\Ncase we have data from experiments or we
Dialogue: 0,0:20:00.90,0:20:06.77,Default,,0000,0000,0000,,use models of calcium oscillations and\Nthen we couple a model of a protein to
Dialogue: 0,0:20:06.77,0:20:13.76,Default,,0000,0000,0000,,these calcium dynamics. This gives us time\Ncourses, both of calcium and protein,
Dialogue: 0,0:20:13.76,0:20:20.26,Default,,0000,0000,0000,,stochastic time courses, including the\Nrandom fluctuations. And then we use the
Dialogue: 0,0:20:20.26,0:20:25.92,Default,,0000,0000,0000,,information-theoretic machinery to study\Nthem. And some of our results I want to show
Dialogue: 0,0:20:25.92,0:20:29.65,Default,,0000,0000,0000,,you. For example, if you increase the\Nsystem size, if you increase the particle
Dialogue: 0,0:20:29.65,0:20:35.48,Default,,0000,0000,0000,,numbers, if you make the cell bigger, then\Nthe information that you can transfer is
Dialogue: 0,0:20:35.48,0:20:40.72,Default,,0000,0000,0000,,higher. Meaning, if the cell invests more\Nenergy and produces more proteins, it can
Dialogue: 0,0:20:40.72,0:20:44.83,Default,,0000,0000,0000,,actually achieve a more reliable\Ninformation transfer, which comes of course
Dialogue: 0,0:20:44.83,0:20:52.12,Default,,0000,0000,0000,,with costs for the cell. Also, it seems,\Nthat if you use more complicated dynamics
Dialogue: 0,0:20:52.12,0:20:56.33,Default,,0000,0000,0000,,- meaning not only spiking, but maybe\Nbursting behavior where you have secondary
Dialogue: 0,0:20:56.33,0:21:00.16,Default,,0000,0000,0000,,spikes - then you can transmit more\Ninformation because the input signal
Dialogue: 0,0:21:00.16,0:21:07.50,Default,,0000,0000,0000,,carries more information or can carry more\Ninformation in its different features.
Dialogue: 0,0:21:07.50,0:21:12.28,Default,,0000,0000,0000,,Another result is that proteins - a very\Ninteresting result I think - is that
Dialogue: 0,0:21:12.28,0:21:17.72,Default,,0000,0000,0000,,proteins can actually be tuned to certain\Ncharacteristics of the calcium input.
Dialogue: 0,0:21:17.72,0:21:22.26,Default,,0000,0000,0000,,Meaning, with all the different calcium\Nsensitive proteins in the cell they are
Dialogue: 0,0:21:22.26,0:21:27.56,Default,,0000,0000,0000,,tuned to a specific signal. So they only\Nget activated or these pathways only
Dialogue: 0,0:21:27.56,0:21:33.27,Default,,0000,0000,0000,,allow information transmission, if a\Ncertain signal is observed in the cell by
Dialogue: 0,0:21:33.27,0:21:39.44,Default,,0000,0000,0000,,these proteins. So, in a way the 3D\Nstructure of the protein defines how it
Dialogue: 0,0:21:39.44,0:21:46.58,Default,,0000,0000,0000,,behaves dynamically, how quickly it binds\Nand so on, how many binding sites it has,
Dialogue: 0,0:21:46.58,0:21:54.62,Default,,0000,0000,0000,,and then this dynamic behavior determines\Nto what input signals that protein is
Dialogue: 0,0:21:54.62,0:21:59.69,Default,,0000,0000,0000,,actually sensitive. On the right hand side\Nyou can see some calculations we did. The
Dialogue: 0,0:21:59.69,0:22:05.62,Default,,0000,0000,0000,,peaks actually show where this specific\Nprotein, which is a calmodulin-like protein
Dialogue: 0,0:22:05.62,0:22:09.78,Default,,0000,0000,0000,,- you don't have to memorize that, it's a\Nvery important calcium sensitive protein -
Dialogue: 0,0:22:09.78,0:22:15.73,Default,,0000,0000,0000,,where these differently parameterized\Nmodels actually get activated and allow
Dialogue: 0,0:22:15.73,0:22:20.49,Default,,0000,0000,0000,,information transfer. And this allows\Ndifferential regulation because you have
Dialogue: 0,0:22:20.49,0:22:25.66,Default,,0000,0000,0000,,all the different proteins. You have only\None calcium concentration and only the
Dialogue: 0,0:22:25.66,0:22:31.68,Default,,0000,0000,0000,,proteins that are sensitive to a specific\Ninput get activated or do their things in
Dialogue: 0,0:22:31.68,0:22:36.21,Default,,0000,0000,0000,,the cell. Now if you look at more\Ncomplicated proteins - so Calmodulin, the
Dialogue: 0,0:22:36.21,0:22:41.24,Default,,0000,0000,0000,,one I just showed you, was only activated\Nby calcium - more complicated proteins,
Dialogue: 0,0:22:41.24,0:22:47.46,Default,,0000,0000,0000,,like protein kinase C, for example, they are\Nboth activated and inhibited. So they show
Dialogue: 0,0:22:47.46,0:22:52.04,Default,,0000,0000,0000,,biphasic behavior, where in an\Nintermediate range of calcium
Dialogue: 0,0:22:52.04,0:22:55.94,Default,,0000,0000,0000,,concentration they get activated, with\Nvery high or very low concentrations they
Dialogue: 0,0:22:55.94,0:23:02.03,Default,,0000,0000,0000,,are inactivated. You can actually see that\Nthese more complicated proteins allow a
Dialogue: 0,0:23:02.03,0:23:07.65,Default,,0000,0000,0000,,higher information transfer and again\Nproducing these more complicated proteins
Dialogue: 0,0:23:07.65,0:23:13.47,Default,,0000,0000,0000,,might be more costly for the cell, but it\Ncan be valuable, because they allow more
Dialogue: 0,0:23:13.47,0:23:18.18,Default,,0000,0000,0000,,information to be transferred. And this\Nyou can see in this plot where we actually
Dialogue: 0,0:23:18.18,0:23:23.09,Default,,0000,0000,0000,,scanned over the activation and the\Ninhibition constant of these model
Dialogue: 0,0:23:23.09,0:23:26.93,Default,,0000,0000,0000,,proteins and you can see that you have\Nthese sweet spots where you get a very
Dialogue: 0,0:23:26.93,0:23:32.10,Default,,0000,0000,0000,,high information transfer. So color coded\Nis transfer entropy. Now, coming to a
Dialogue: 0,0:23:32.10,0:23:37.63,Default,,0000,0000,0000,,different system: Just quickly, we also\Nlooked at other systems of course. Calcium
Dialogue: 0,0:23:37.63,0:23:42.56,Default,,0000,0000,0000,,signaling is just one of our favorite ones.\NWe also looked at bacteria and this is
Dialogue: 0,0:23:42.56,0:23:50.58,Default,,0000,0000,0000,,E. coli, a very famous model system for\Nbiologists. These are cells that can
Dialogue: 0,0:23:50.58,0:23:58.62,Default,,0000,0000,0000,,actually move around because they have\Nlittle propellers at their end. They want to
Dialogue: 0,0:23:58.62,0:24:05.10,Default,,0000,0000,0000,,find sources of nutrients, for example, to\Nget food. So they swim into a direction
Dialogue: 0,0:24:05.10,0:24:10.91,Default,,0000,0000,0000,,and then they decide whether\Nto keep swimming in that direction
Dialogue: 0,0:24:10.91,0:24:17.34,Default,,0000,0000,0000,,or whether to tumble, reorient randomly,\Nand swim in some other direction. The
Dialogue: 0,0:24:17.34,0:24:24.11,Default,,0000,0000,0000,,problem for them is they are too small.\NThey can't detect a concentration gradient
Dialogue: 0,0:24:24.11,0:24:30.04,Default,,0000,0000,0000,,of nutrients, of food between their front\Nand the back of the cell. So they have to
Dialogue: 0,0:24:30.04,0:24:35.48,Default,,0000,0000,0000,,swim in one direction and then they have\Nto remember some nutrient concentration of
Dialogue: 0,0:24:35.48,0:24:40.54,Default,,0000,0000,0000,,some time back and then they have to\Ncompare: Is the nutrient
Dialogue: 0,0:24:40.54,0:24:44.72,Default,,0000,0000,0000,,concentration actually increasing? Then I\Nshould continue swimming. If it's
Dialogue: 0,0:24:44.72,0:24:49.67,Default,,0000,0000,0000,,decreasing, I should reorient and swim in\Nsome other direction. This allows them to,
Dialogue: 0,0:24:49.67,0:24:58.28,Default,,0000,0000,0000,,on average, swim towards sources of food.\NIn order to compare over time the nutrient
Dialogue: 0,0:24:58.28,0:25:04.75,Default,,0000,0000,0000,,concentrations they have to memorize, they\Nhave to know how much nutrients where
Dialogue: 0,0:25:04.75,0:25:11.73,Default,,0000,0000,0000,,there sometime ago. For that they have a\Nlittle memory and the memory is actually
Dialogue: 0,0:25:11.73,0:25:16.74,Default,,0000,0000,0000,,in the - you can see on the left hand side\Nthe receptor that actually senses these
Dialogue: 0,0:25:16.74,0:25:21.96,Default,,0000,0000,0000,,nutrients. They can be modified, these\Nreceptors, we call that methylated. So they
Dialogue: 0,0:25:21.96,0:25:27.13,Default,,0000,0000,0000,,get a methylation group attached. They\Nhave different states of methylation, five
Dialogue: 0,0:25:27.13,0:25:33.84,Default,,0000,0000,0000,,different ones in that model we are\Nlooking at. This builds a memory. And we
Dialogue: 0,0:25:33.84,0:25:38.40,Default,,0000,0000,0000,,looked into that, we quantified that with\Ninformation theory. This is a measure,
Dialogue: 0,0:25:38.40,0:25:42.80,Default,,0000,0000,0000,,this is called mutual information. It's\Nnot transfer entropy, it's another measure
Dialogue: 0,0:25:42.80,0:25:50.06,Default,,0000,0000,0000,,of, in that case, statical information.\NYou can see, this is the amount of
Dialogue: 0,0:25:50.06,0:25:55.63,Default,,0000,0000,0000,,information that is actually stored about\Nthe nutrient concentration that is outside
Dialogue: 0,0:25:55.63,0:26:01.99,Default,,0000,0000,0000,,of the cell. This is in nats, it's not in\Nbits. It's just a different - you can
Dialogue: 0,0:26:01.99,0:26:06.94,Default,,0000,0000,0000,,translate them - it's just a different unit\Nfor information. You can also see how the
Dialogue: 0,0:26:06.94,0:26:14.23,Default,,0000,0000,0000,,different methylation states - so these\Nare the colored curves - how they go
Dialogue: 0,0:26:14.23,0:26:22.23,Default,,0000,0000,0000,,through or how they are active with\Ndifferent nutrient concentrations. This is
Dialogue: 0,0:26:22.23,0:26:26.33,Default,,0000,0000,0000,,ongoing research. So, maybe, next time, \Nhopefully, next time, I can show you much
Dialogue: 0,0:26:26.33,0:26:32.29,Default,,0000,0000,0000,,more. Just to finish this, we also look at\Ntimescales, because the timescales have to
Dialogue: 0,0:26:32.29,0:26:38.76,Default,,0000,0000,0000,,be right. The system adapts. So if you\Nkeep that cell in a certain nutrient
Dialogue: 0,0:26:38.76,0:26:42.74,Default,,0000,0000,0000,,concentration, it adapts to that nutrient\Nconcentration and goes back to its normal
Dialogue: 0,0:26:42.74,0:26:48.43,Default,,0000,0000,0000,,operating level. Now, if you increase the\Nnutrient concentration again, it shows some
Dialogue: 0,0:26:48.43,0:26:53.95,Default,,0000,0000,0000,,swimming behavior. So it adapts, but it\Nalso has to decide, it also has to compare
Dialogue: 0,0:26:53.95,0:26:59.62,Default,,0000,0000,0000,,the different nutrients at different\Npositions. That's how they have to manage
Dialogue: 0,0:26:59.62,0:27:04.68,Default,,0000,0000,0000,,the different timescales of decision\Nmaking and memory or adaptation and we are
Dialogue: 0,0:27:04.68,0:27:10.12,Default,,0000,0000,0000,,looking into that as well. Coming to the\Nconclusions, I hope I could convince you
Dialogue: 0,0:27:10.12,0:27:14.15,Default,,0000,0000,0000,,that information theory can be applied to\Nbiology, that it's a very interesting
Dialogue: 0,0:27:14.15,0:27:22.72,Default,,0000,0000,0000,,topic, it's a fascinating area and we are\Njust at the beginning to do that. I also
Dialogue: 0,0:27:22.72,0:27:28.97,Default,,0000,0000,0000,,showed you that it's such that in\Nsignaling pathways the components can be
Dialogue: 0,0:27:28.97,0:27:34.29,Default,,0000,0000,0000,,tuned to their input, which allows\Ndifferential regulation. So even though
Dialogue: 0,0:27:34.29,0:27:40.23,Default,,0000,0000,0000,,you don't have wires you can still\Nspecifically activate different proteins
Dialogue: 0,0:27:40.23,0:27:49.82,Default,,0000,0000,0000,,with one signal or multiplex, if you want.\NWe are of course in the process of
Dialogue: 0,0:27:49.82,0:27:55.97,Default,,0000,0000,0000,,studying what features of the input signal\Nare actually information-carrying. So we
Dialogue: 0,0:27:55.97,0:28:02.99,Default,,0000,0000,0000,,are looking into things like wave form and\Ntiming. And we want to look into how these
Dialogue: 0,0:28:02.99,0:28:08.73,Default,,0000,0000,0000,,things change in the deceased case. So, if\Nyou have things like cancer where certain
Dialogue: 0,0:28:08.73,0:28:15.61,Default,,0000,0000,0000,,signalling pathways are perturbed or fail,\Nwe want to exactly find out what does that
Dialogue: 0,0:28:15.61,0:28:21.27,Default,,0000,0000,0000,,do to the information processing\Ncapabilities of the cell. We also found
Dialogue: 0,0:28:21.27,0:28:27.32,Default,,0000,0000,0000,,out that estimating these information\Ntheoretical quantities can be a very
Dialogue: 0,0:28:27.32,0:28:33.39,Default,,0000,0000,0000,,tricky business. Another project we are\Ndoing at the moment is actually only on
Dialogue: 0,0:28:33.39,0:28:39.93,Default,,0000,0000,0000,,how to interpret these in a reliable\Nmanner, how to estimate these from sparse
Dialogue: 0,0:28:39.93,0:28:45.28,Default,,0000,0000,0000,,and noisy data. So that's also ongoing\Nwork. I would like to thank some of my
Dialogue: 0,0:28:45.28,0:28:50.96,Default,,0000,0000,0000,,collaborators, of course, my own group, but\Nalso some others, in particular the Copasi
Dialogue: 0,0:28:50.96,0:28:57.40,Default,,0000,0000,0000,,team, that is spread all over the world.\NAnd with that I would like to thank you
Dialogue: 0,0:28:57.40,0:29:00.94,Default,,0000,0000,0000,,for your attention and I would be happy to\Nanswer any question you might have.
Dialogue: 0,0:29:00.94,0:29:02.00,Default,,0000,0000,0000,,Thank you.
Dialogue: 0,0:29:02.00,0:29:05.21,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:29:05.21,0:29:13.05,Default,,0000,0000,0000,,Herald Angel: ... a very warm applause\Nfor Jürgen. If you have questions, there
Dialogue: 0,0:29:13.05,0:29:16.38,Default,,0000,0000,0000,,are two microphones, microphone number\None, microphone number two and please
Dialogue: 0,0:29:16.38,0:29:23.18,Default,,0000,0000,0000,,speak loudly into the microphone. And, I think\Nthe first one is microphone number two.
Dialogue: 0,0:29:23.18,0:29:25.06,Default,,0000,0000,0000,,Your question please.\NMicrophone 2: Has there been any work done
Dialogue: 0,0:29:25.06,0:29:29.53,Default,,0000,0000,0000,,on computational modelling of the G-protein\Ncoupled receptors and the second messenger
Dialogue: 0,0:29:29.53,0:29:32.45,Default,,0000,0000,0000,,cascades there.\NJürgen: Can you repeat that, sorry.
Dialogue: 0,0:29:32.45,0:29:36.18,Default,,0000,0000,0000,,Microphone 2: Has there any work been done\Non computational modelling of G protein-
Dialogue: 0,0:29:36.18,0:29:37.91,Default,,0000,0000,0000,,coupled receptors\NJürgen: G protein?
Dialogue: 0,0:29:37.91,0:29:40.11,Default,,0000,0000,0000,,Microphone 2: Yeah.\NJürgen: Oh yes, I mean we are doing that
Dialogue: 0,0:29:40.11,0:29:44.46,Default,,0000,0000,0000,,because calcium is actually... I mean the\Ncalcium signal is actually triggered by a
Dialogue: 0,0:29:44.46,0:29:49.58,Default,,0000,0000,0000,,cascade that includes the G protein. Most\Nof these receptors are actually G coupled
Dialogue: 0,0:29:49.58,0:29:54.41,Default,,0000,0000,0000,,or G protein coupled receptors. So that's\Nwhat we are doing.
Dialogue: 0,0:29:54.41,0:29:57.50,Default,,0000,0000,0000,,Angel: Thank you. Microphone number two\Nagain.
Dialogue: 0,0:29:57.50,0:30:01.46,Default,,0000,0000,0000,,Microphone 2: First of all thanks for the\Ntalk. I want to ask you talked a
Dialogue: 0,0:30:01.46,0:30:07.71,Default,,0000,0000,0000,,little bit about how different proteins\Nget activated by different signals and
Dialogue: 0,0:30:07.71,0:30:15.19,Default,,0000,0000,0000,,could you go a bit into detail about what\Nkind of signal qualities the proteins can
Dialogue: 0,0:30:15.19,0:30:22.31,Default,,0000,0000,0000,,detect? So are they triggered by specific\Nfrequencies or specific decays, like which
Dialogue: 0,0:30:22.31,0:30:27.100,Default,,0000,0000,0000,,characteristics of the signals can be\Npicked up by the different proteins?
Dialogue: 0,0:30:27.100,0:30:32.43,Default,,0000,0000,0000,,Jürgen: Well, that's actually what we\Nstudy. I mean we have another package that
Dialogue: 0,0:30:32.43,0:30:36.87,Default,,0000,0000,0000,,is linked here, the last one, the\Noscillator generator. This is a package in
Dialogue: 0,0:30:36.87,0:30:43.18,Default,,0000,0000,0000,,R that allows you to create artificial\Ninputs, where you have complete control of
Dialogue: 0,0:30:43.18,0:30:48.81,Default,,0000,0000,0000,,all the parameters like amplitude and\Nduration of the peak, duration of the
Dialogue: 0,0:30:48.81,0:30:55.23,Default,,0000,0000,0000,,secondary peak, frequencies of the primary\Npeaks of the secondary peaks, refraction
Dialogue: 0,0:30:55.23,0:30:59.55,Default,,0000,0000,0000,,period and so on. You have complete\Ncontrol and at the moment we are also
Dialogue: 0,0:30:59.55,0:31:04.68,Default,,0000,0000,0000,,running scans and want to find out what\Nproteins are actually sensitive to what
Dialogue: 0,0:31:04.68,0:31:09.90,Default,,0000,0000,0000,,parameters in the input signal. What we\Nknow from calcium is that, for example,
Dialogue: 0,0:31:09.90,0:31:18.60,Default,,0000,0000,0000,,calcium calmodulin kinase 2, also a very\Nimportant protein in the nervous system,
Dialogue: 0,0:31:18.60,0:31:25.77,Default,,0000,0000,0000,,that shows frequency modulation. It has\Nalso been shown experimentally where they
Dialogue: 0,0:31:25.77,0:31:29.70,Default,,0000,0000,0000,,put that protein on a surface, they\Nimmobilized it on a surface, and then they
Dialogue: 0,0:31:29.70,0:31:34.49,Default,,0000,0000,0000,,superfused it with calcium concentrations\Nor with solutions of different calcium
Dialogue: 0,0:31:34.49,0:31:39.15,Default,,0000,0000,0000,,concentration in a pulsed manner and they\Nmeasured the activity of that protein and
Dialogue: 0,0:31:39.15,0:31:44.72,Default,,0000,0000,0000,,they showed that, with increasing frequency,\Nthe activation gets bigger. At the same
Dialogue: 0,0:31:44.72,0:31:48.32,Default,,0000,0000,0000,,time it also shows amplitude modulation,\Nokay? It's also sensitive to the
Dialogue: 0,0:31:48.32,0:31:55.25,Default,,0000,0000,0000,,amplitude, meaning the absolute height of\Nthe concentration of calcium.
Dialogue: 0,0:31:55.25,0:31:57.11,Default,,0000,0000,0000,,Microphone 2: Thanks.\NJürgen: Thank you.
Dialogue: 0,0:31:57.11,0:32:00.72,Default,,0000,0000,0000,,Angel: And again number two please.\NMicrophone 2: Hey. So you talked about a
Dialogue: 0,0:32:00.72,0:32:07.17,Default,,0000,0000,0000,,lot of on and off kinetics and I wonder, if\Nyou think about neurons, which are not only
Dialogue: 0,0:32:07.17,0:32:14.16,Default,,0000,0000,0000,,having on and off, but also many amplitudes\Nthat take a big role in development of
Dialogue: 0,0:32:14.16,0:32:20.99,Default,,0000,0000,0000,,cells and synapses. How do you measure\Nthat, so how do you measure like baseline,
Dialogue: 0,0:32:20.99,0:32:26.04,Default,,0000,0000,0000,,sporadic activity of calcium?\NJürgen: Well, in our case there are
Dialogue: 0,0:32:26.04,0:32:28.99,Default,,0000,0000,0000,,different ways of measuring calcium.\NThat's not what we are doing...
Dialogue: 0,0:32:28.99,0:32:32.46,Default,,0000,0000,0000,,Microphone 2: ... not really measuring,\Nsorry, but more like how do you integrate it
Dialogue: 0,0:32:32.46,0:32:37.42,Default,,0000,0000,0000,,in your system? Because it's not really an\Non/off reaction but it's more like a
Dialogue: 0,0:32:37.42,0:32:43.45,Default,,0000,0000,0000,,sporadic miniature.\NJürgen: Yeah, I mean in the case of
Dialogue: 0,0:32:43.45,0:32:48.92,Default,,0000,0000,0000,,calcium you have these time courses, okay?\NAnd we look at the complete time
Dialogue: 0,0:32:48.92,0:32:53.18,Default,,0000,0000,0000,,course. So we have the calcium\Nconcentration sampled at every second or
Dialogue: 0,0:32:53.18,0:32:58.62,Default,,0000,0000,0000,,half second in the cell by different\Nmethods. So our collaboration partners
Dialogue: 0,0:32:58.62,0:33:04.92,Default,,0000,0000,0000,,they use different dyes that show\Nfluorescence, say, when they bind calcium.
Dialogue: 0,0:33:04.92,0:33:10.25,Default,,0000,0000,0000,,Some others show bioluminescence. And then\Nwe use these time courses. In the neural
Dialogue: 0,0:33:10.25,0:33:17.82,Default,,0000,0000,0000,,system it's a bit different. There you\Nalso get the analog mode, where neurons are
Dialogue: 0,0:33:17.82,0:33:23.60,Default,,0000,0000,0000,,directly connected and they exchange\Nsubstances, but most of the case you have
Dialogue: 0,0:33:23.60,0:33:28.73,Default,,0000,0000,0000,,action potentials and I didn't go into\Nneural systems at all because things there
Dialogue: 0,0:33:28.73,0:33:34.44,Default,,0000,0000,0000,,are totally different. You get these\Naction potentials that are uniform mostly,
Dialogue: 0,0:33:34.44,0:33:38.10,Default,,0000,0000,0000,,so they they all have the same duration,\Nthey all have the same amplitude. And then
Dialogue: 0,0:33:38.10,0:33:44.06,Default,,0000,0000,0000,,people in neuroscience or computational\Nneuroscience mostly they boil the
Dialogue: 0,0:33:44.06,0:33:49.97,Default,,0000,0000,0000,,information down to just the timings of\Nthese peaks and they use this information
Dialogue: 0,0:33:49.97,0:33:53.94,Default,,0000,0000,0000,,and mathematically this is a point process\Nand you can use different mathematical
Dialogue: 0,0:33:53.94,0:33:59.83,Default,,0000,0000,0000,,tools to study that. We are not really\Nlooking into neurons. We are mostly
Dialogue: 0,0:33:59.83,0:34:06.56,Default,,0000,0000,0000,,interested in non-excitable cells, like\Nliver cells, pancreatic cells and so on,
Dialogue: 0,0:34:06.56,0:34:12.32,Default,,0000,0000,0000,,cells that are not activated, they don't\Nshow massive depolarization, like
Dialogue: 0,0:34:12.32,0:34:18.08,Default,,0000,0000,0000,,neurons. Thank you.\NAngel: Thank you. And obviously again
Dialogue: 0,0:34:18.08,0:34:22.23,Default,,0000,0000,0000,,number two.\NMicrophone 2: Hi. So, you mentioned CaM
Dialogue: 0,0:34:22.23,0:34:28.70,Default,,0000,0000,0000,,kinases 2. I got that you don't work\Non neuroscience specifically, but I'm
Dialogue: 0,0:34:28.70,0:34:33.46,Default,,0000,0000,0000,,pretty sure you have a quite extensive\Nknowledge in the subject. What do you
Dialogue: 0,0:34:33.46,0:34:42.07,Default,,0000,0000,0000,,think about this, I would say, hypotheses\Nthat were quite popular a few years ago, I
Dialogue: 0,0:34:42.07,0:34:48.66,Default,,0000,0000,0000,,think in the US mainly, about the fact\Nthat the cytoskeleton of neurons can
Dialogue: 0,0:34:48.66,0:34:58.08,Default,,0000,0000,0000,,actually encode and decode through kinases\Nin the cytoskeleton memories like bits in
Dialogue: 0,0:34:58.08,0:35:03.21,Default,,0000,0000,0000,,- you know - in a hard drive. What's your\Nfeeling?
Dialogue: 0,0:35:03.21,0:35:07.21,Default,,0000,0000,0000,,Jürgen: Well, I'm not going to speculate\Non that specific hypothesis because I'm
Dialogue: 0,0:35:07.21,0:35:12.32,Default,,0000,0000,0000,,not really into that, but I know that many\Npeople are also looking into spatial
Dialogue: 0,0:35:12.32,0:35:16.71,Default,,0000,0000,0000,,effects which I didn't mention here. I\Nmean the model I showed you is a spatially
Dialogue: 0,0:35:16.71,0:35:22.26,Default,,0000,0000,0000,,homogeneous model. We don't look at\Nconcentration gradients within the cell,
Dialogue: 0,0:35:22.26,0:35:27.14,Default,,0000,0000,0000,,our cells are homogeneous at the moment,\Nbut people do that. And of course then you
Dialogue: 0,0:35:27.14,0:35:33.50,Default,,0000,0000,0000,,can look into things, for example, like a\Nnew topic is morphological computation,
Dialogue: 0,0:35:33.50,0:35:39.08,Default,,0000,0000,0000,,meaning that spatially you can also\Nperform computations. But, if you're
Dialogue: 0,0:35:39.08,0:35:41.12,Default,,0000,0000,0000,,interested in that, I mean, we can\Ntalk offline...
Dialogue: 0,0:35:41.12,0:35:43.56,Default,,0000,0000,0000,,Microphone 2: ... do you buy into this\Ntheory...
Dialogue: 0,0:35:43.56,0:35:45.39,Default,,0000,0000,0000,,Jürgen: ... I can give you some pointers\Nthere..
Dialogue: 0,0:35:45.39,0:35:49.53,Default,,0000,0000,0000,,Microphone 2: ... but do you have a good\Nfeeling about these theories or you think
Dialogue: 0,0:35:49.53,0:35:52.02,Default,,0000,0000,0000,,they're clueless.\NJürgen: Well, I think that the spatial
Dialogue: 0,0:35:52.02,0:35:56.06,Default,,0000,0000,0000,,aspect is a very important thing. And\Nthat's also something we should
Dialogue: 0,0:35:56.06,0:36:01.61,Default,,0000,0000,0000,,look at. I mean, to me random fluctuations\Nare very important, intrinsic fluctuations
Dialogue: 0,0:36:01.61,0:36:05.82,Default,,0000,0000,0000,,because you can't separate them from the\Ndynamics of the system. They are always
Dialogue: 0,0:36:05.82,0:36:11.59,Default,,0000,0000,0000,,there, at least some of the fluctuations.\NAnd also the spatial effects are very
Dialogue: 0,0:36:11.59,0:36:15.23,Default,,0000,0000,0000,,important, because you not only\Nhave these different compartments,
Dialogue: 0,0:36:15.23,0:36:19.98,Default,,0000,0000,0000,,where the reactions happen, but you also\Nhave concentration gradients across the
Dialogue: 0,0:36:19.98,0:36:24.55,Default,,0000,0000,0000,,cell. Especially with calcium, people have\Nlooked into calcium puffs and calcium
Dialogue: 0,0:36:24.55,0:36:29.77,Default,,0000,0000,0000,,waves because, when you have a channel, that\Nallows calcium to enter, of course directly
Dialogue: 0,0:36:29.77,0:36:34.06,Default,,0000,0000,0000,,at that channel you get a much higher\Ncalcium concentration and then in some
Dialogue: 0,0:36:34.06,0:36:39.71,Default,,0000,0000,0000,,cases you get waves that are travelling\Nacross the the cell. And to me it sounds
Dialogue: 0,0:36:39.71,0:36:43.99,Default,,0000,0000,0000,,plausible that this also has a major\Nimpact on the information processing.
Dialogue: 0,0:36:43.99,0:36:49.39,Default,,0000,0000,0000,,Yeah. Thank you.\NAngel: Thank you. In this case, Jürgen,
Dialogue: 0,0:36:49.39,0:36:54.54,Default,,0000,0000,0000,,thank you for the talk. And please give a\Nvery warm applause to him.
Dialogue: 0,0:36:54.54,0:36:56.14,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:36:56.14,0:36:58.61,Default,,0000,0000,0000,,Jürgen: Thank you.
Dialogue: 0,0:36:58.61,0:37:03.43,Default,,0000,0000,0000,,{\i1}applause{\i0}
Dialogue: 0,0:37:03.43,0:37:08.32,Default,,0000,0000,0000,,{\i1}postroll music{\i0}
Dialogue: 0,0:37:08.32,0:37:26.00,Default,,0000,0000,0000,,subtitles created by c3subtitles.de\Nin the year 2019. Join, and help us!