 Hello and welcome back beyond networks the evolution of living systems. I hope you've enjoyed my rant about biology really being systems biology there's nothing beyond systems biology all of biology is about systems in some way that's exactly what what sets it apart from mere physics or chemistry or biochemistry so that's all good and fine but if we think that biology should be systems biology then we also need to have a better idea of what a system is everybody's going around these days saying they're doing systems biology all the genomics people you sequence a few genomes and you do systems biology you do some transcriptomics you have some network graphs somewhere everybody in their grand application is doing systems biology this lecture is part of a master's in evolutionary systems biology there's molecular systems biology I don't know how that works and even the European molecular biology organization and Institute have become focused on systems biology although molecular biology used to be exactly the opposite of systems biology you would go in and study organisms at the molecular level so you would basically get rid of the organism like we've seen in the last lecture so how can we make biology look at organisms again we have to think about systems it requires a slightly different thinking but before we can even talk about what that is we need to think a little bit about what a system really is so let's spend a few minutes defining a system and let's go to sort of some sort of random online dictionary Merriam-Webster in this case and look up the definition Merriam-Webster dictionary tells us that a system is a regularly interactive interacting or independent group of items forming a unified there's already something really interesting here so a group of items they are interdependent they interact so it's dynamic and they form a unified whole we have to think a lot about what that means what when is the system a whole and what is that so this is very concise very short very precise but not very useful so let's expand it a little bit this is from a paper by Hall and Fagan in 1956 channeling the views of Ludwig von Bertalanffi the Viennese godfather of modern systems theory and they say a system is a set of objects together with the relationships between the objects and between their attributes so there are relationships their objects this definition is sort of it's a bit static for me right it's relational it tries to relate different different objects together but it doesn't really do it yet for me so let's go on and look and add to this definition here's Michael Savageau a biochemist and modeler who says a system can be defined as a collection of interacting parts so here's the clockwork universe underneath again which in some sense constitutes a whole that's no longer the clockwork universe what is that whole everything excluded from the collection is considered the environment of the system oh okay so the definition of a system is very important is setting a system apart from the rest of the universe so there must be a boundary around the system so if we want the truth and the full truth of course we need to go to Wikipedia and look at what it has to say about systems and here's the definition from the current Wikipedia article on the topic and it says a system is a group of interacting or interrelated entities that form a unified whole so far so good we've had this already the system is described by its spatial and temporal boundaries okay surrounded and influenced by its environment which is outside the boundaries it's described by its structure and purpose and expressed in its functioning okay we need to take this apart a little bit and then systems are the subjects of study of systems theory fair enough so this is what we're gonna try and do we're gonna apply some sort of very philosophical systems theory here to biology in this lecture so let's examine this a little more so what are the entities that are interacting or interrelated here we have to think really hard about that so there's several questions involved here so the entities that are interacting could be objects they could be events they could be agents they could be processes it's not clear what they are so we need to think a little bit more about that you may have a hunch what I prefer in this context also the other question is if we have a system are the entities that interact specific examples of something or are they whole classes of things we can define a system over more than just a specific instance okay so are these entities in philosophy speak we would say tokens or classes okay that's a question that will keep us busy as well let's go on and say a system is described by its spatial and temporal boundaries there's a really big question here and how do we define these boundaries especially if you tend towards being interested in systems that are made of processes I told you already when we introduced process thinking that one of the major drawbacks of a process perspective is that it's really difficult to define the exact spatial temporal boundaries of a process this will become a really big problem and the biggest sort of mystery in here in this quote is how does the structure of a system relate to its function or purpose so it is quite obvious what the structure is some sort of we draw the components and their interactions and we get this famous network graph okay they have a representation of the structure of the system it's also sometimes called the topology of a system and so here's the network metaphor and that's very useful and that's also very straightforward to describe you you have to just know and we'll give you an example of this in a second what the components are that you're interested in and how they interact okay but how do you get from there to the function of a system what is the function of the system we'll ask that question again later on and then the purpose so his this function is always somehow linked to a purpose of the system if the system is for something and that of course is a very controversial notion in biology which we have to discuss in detail so let's expand let's move on from these very simple definitions of systems to a bit more realistic dynamic and also complex definition which comes from an absolutely wonderful book Richard Luntin's and Levin's biology under the influence a collection of absolutely fantastic essays and one of those essays is called educating the intuition to cope with complexity I highly recommend you read that and the major statement it makes is that very Wimsatian sort of view of the world Wimsat by the way was Luntin student and in this essay they argue the authors Levin's and Luntin argued that complexity is measured by the number of different valid perspectives basically that you can have on the system they don't quite put it in those words but they also give a very interesting definition of what a system is let me quote that a system is a network of variables okay this is interesting so variables implied at the entities that are involved in a system they can change over time but variables are also something that we use in mathematics it's a formal concept to describe something real a phenomenon that's happened so we'll have to think about what is the correlation between a sort of a formal system and an actual system out there in the world so these variables are linked by positive and negative feedbacks I like that we haven't had that before so if the whole is to be unified what does that mean and here we're starting to think about that when does it make sense to to talk about a whole at the system level okay many systems are just aggregates of parts these parts interact and their collective behavior is just additive you can add up their individual contributions and you get an aggregate for example the atoms that contribute to the rock you know they form granite all together they all have the same contribution and you can just add them up and you get in the end from the molecules you get a rock there's no mystery there there's no is there a whole is there a rock okay these are questions we have to discuss but in systems where you have lots of positive and negative feedback loops this becomes very different and it becomes very different difficult to predict the behavior of the system at its systems level from the behavior of the component parts and in addition to this sort of feedback these variables are usually not in equilibrium but in continual movement within limits and around an equilibrium state so they're sort of influenced by what we will later in this lecture called called attractors of a dynamical system but they're not quite at the attractor which is a state of studies is a steady state it's a sort of a form of dynamic equilibrium so they're far from that okay and they're constantly in flux further each part has its own dynamics so here is Simon's near decomposability again or modularity of systems how it responds to outside impacts so there's a sensory sort of function systems perceive input from outside and erases those impacts each at its own rate so in a way they also implied as a very rich quote here they imply that the behavior of the system depends on its history and each module of the system has a different sort of memory of that history so this is sort of a bridge into this sort of classic definition of a complex adaptive system that goes back to work at the Santa Fe Institute by John Holland and Murray Gell-Mann and others at the time and they came up with a sort of a formal definition of what a complex adaptive system is that's very widely used right now and here's a sort of a cartoon version of a complex system I've chosen a mouse a rodent and its prey here and you can see that there are there's some sort of on this side of the graph there's some sort of input their sensory input the system is perceiving its environment there's some food some energy that goes into the system the system has a very complicated structure components that are interacting just like we had in these in these very simple definitions of systems and they constitute the mouse here which as a whole as a unified whole constrains the behavior of its parts will come back that's very important so there's there's not only a sort of a causal flow up from the components that constitute the whole system but the behavior of the whole system which has some coherence influences and construct by constraining for example the behavior of its parts so there's there's a feedback not just among components of the the molecular sort of subsystems down here or the cellular subsystems but there's a feedback between the systems level and the component level below it and so you get from this very complex of the energy flowing through this this complex structured rule-based sort of system you get some sort of complex behavior like cheese eating and gathering here we can very you know list a few more characteristics and make this division of definition of a complex adaptive system a bit more precise so a complex adaptive system there's a very important distinction you can have a complicated system or you can have a complex system a complex system shows systems level behavior that's not easy to predict from its components a complicated system is just that a lot of different interactions it's a big mess but it doesn't necessarily show this independent system level behavior it can be a pure complicated aggregate of interactions so here we have a large number of components and they have to interact in nonlinear way ways because linear systems are always predictable and aggregate the structure of the system this is how I've drawn this network has to be multi-level it's hierarchic hierarchical and mouses is made out of tissues which are made out of cells organs tissue cells the cells are made out of biochemical macromolecules genes proteins and so on and so forth which are made out of molecules and atoms of course okay so we have a multi-level structure and that structure is near decomposable I've also used a network that has a modular structure with hubs so there are certain nodes like the central one in the network I've drawn here that are connected to a lot of other hubs other nodes in the network okay so what you get is called a small world topology which means that if you want to get from one node to the other if you have such central hubs in the network you can get very quickly from any one node to any other node this is called small world structure small world topology and also the sort of number of connections that a node has in a network which is called its degree distribution the degree distribution of the network so different nodes of different numbers of connections and this distribution in the case of complex adaptive systems typically follows what is called a parallel which is a distribution with a very fat tail Nassim Taleb loves to talk about fat tails fat tails are statistical events or statistical elements like nodes in the network that are exceptional but very important like these hubs in the network are very rare in the one I've drawn here there's only one central hub of all the nodes but they have a disproportionately big influence on the network and also the evolution of the system so these are these are sort of technical term terms that I would like you to remember they're going to be very important in the future also very importantly any complex adaptive system is far from thermodynamic equilibrium and it's open that means it's open to flows of matter and energy if you don't eat you die if a plant doesn't get any sunlight it dies and this is an essential sort of aspect of living to come back to that and also the system has to be structurally coupled with the environment it perceives its environment but it also influences its environment in unexpected ways to come back to that later and so in this sense there's a lot of positive and negative feedback not only between components of the molecular or the subnetwork level but also between the system as a whole and the level of its components they influence each other reciprocally so this is where reductionism obviously fails because it assumes that all the behavior can be reconstructed from decomposing the systems the system into its component parts okay so dynamics are history or path-dependent so the system complex adaptive systems need memory otherwise they cannot adapt and complex global behavior emerges from local interactions but the whole is more than the sum of its parts otherwise it's just an aggregate and I've said this before lastly and and maybe the most important feature of complex adaptive systems is there they're not just resilient or robust against perturbations they're anti-fragile that means that they're adapted that's the very definition of it at activity they learn from errors in an evolutionary sense they learn a population learns from some of its individuals dying but also complex organisms themselves have adaptive behavior they can adapt their behavior within a lifetime to their environmental conditions and on an evolutionary scale that leads to some of the components being generatively entrenched we've talked about this before so when components become very important so that if you would remove them too much of the system would crumble with them they cannot be removed anymore and they become an essential part of that system while others that are more at the periphery and don't have so many dependencies can change and this will greatly influence the evolutionary dynamics of the system just to remind you as well so I talked about organisms here right and we're going to continue talking about organisms but as I said in my lecture about Bill Wimsatz's perspectivism you can apply this abstract formal definition of a complex adaptive system to science itself and its institutions and its communities instead of food and perception you get smart people energy and funding going into the system the people that are in the system constitute the system but it constrains our career choices our research questions and so you get in the end a complex scientific theories out of this very complex adaptive systems behavior and in the end you get a scientific worldview that is able to adapt just like an organism is adapting to its environment our scientific theories are adapting to our current environment and the problems we are encountering in that environment so let me give you let me sort of come back from this sort of complex systems level to to give you a very very simple example of how what it means to define a system and what the difficulties are that are involved so let's introduce for the first time in this lecture my favorite animal which is not a fruit fly fruit flies are tephra did flies that are parasites on fruit this is a vinegar fly it's called Drosophila melanogaster and it doesn't eat fruit here it's standing on a banana leaf but it's actually eating microorganisms that grow on decaying fruit so it's not an agricultural pest at all it has lovely red eyes a segmented body plan which you can see here very clearly in the abdomen but also by the arrangements of its legs and for a for a big large time of my career I've spent trying to model and study the gene regulatory networks that lead to the formation of this beautiful segmented body plan and there are a bunch of genes that are involved in that process which are called the gap genes so we're going to look at those and what I'm showing you here is a Drosophila embryo this ball is one huge cell about between a third and a half a millimeter long so visible by the naked eye just one cell with many many many nuclei every dot you can see here in this picture is a single nucleus but these nuclear are not separated by cell membranes yet there's just one cell membrane around the whole thing and here's the future head end on the left and the future tail end is on the right and this embryo has been peeled the technical term is decoriented and then colored for two different gene products so what you see here is the distribution of two proteins that are encoded by gap genes they are transcription factor that right factors that regulate other genes expression and they are present in only a subset of the nuclear in the embryo you can see in blue is a gap gene called giant and in green is a protein made by a gap gene called crouple and so these two genes they regulate each other they are transcription factors so what we want to find out is how do they regulate each other and how do they generate the pattern that we see in this embryo this is what we want to know and so we want to define a system which in this case is easy we can just say okay here we want to know how these genes interact so let's just take them and their interactions and let that be the system so here we're trying to go from those genes and the proteins they make they go through the embryo and they bind to the other gene the genes are represented by these boxes the arrows indicate where the DNA sequence becomes transcribed in these transcription factors they bind to each other and for example the the green transcription factor goes and binds here in front of the blue gene and so now the question is how do these two genes interact to create the pattern that we see in the embryo and the striking feature of this embryo of course is that none of the nuclei express both of them at the same time so basically what you have here is a very strong mutual inhibition of these two factors which I represent by these two t-bars that make up this mini tiny gene network here so you have two genes they interact each other with each other like this of course there are other factors for example those factors that have to activate them in the first place they're not considered here yet and if you put that simple system and this negative double negative feedback loop which is a positive feedback loop twice negative positive right you inhibit an inhibitor that's an activation so that forms a positive feedback loop that locks the cells that carry this network either in the green state or the blue state but now we can ask ourselves what have we done here okay we can ask ourselves a whole series of questions how did we define the system why did we choose those two genes or gene products I didn't tell you I gave this example because it's visually appealing and it's very simple okay but there why would you draw a boundary there I can tell you that there are other gap genes that also regulate these genes so it was completely arbitrary in a way but for my purpose at this point it was exactly what I wanted I only wanted to talk about these two genes because I wanted to have an example that's very simple so for my purpose this way of defining the system was completely okay but then if you're if you're actually more serious about studying this pattern forming process you have to think about why do we focus on transcriptional regulation only there are many other biochemical regulatory processes going on the RNAs of these genes are spliced there is trans post transcriptional and post translational regulation even those proteins are modified with different phosphorylation etc and their activity is affected in this way so I have to justify at some point why did I focus on only one step in this entirely obscure and very complicated biochemical cascade I focus on genetic interactions only I can tell you that the molecular details of these proteins binding to the other genes DNA sequence are very very complicated but I've just ignored all of them and said okay so there's just a net positive or negative in this case negative interaction repressive interaction between the two genes what allows me to choose this level of abstraction I and idealize away all the details the molecular details of this process why did I only choose a specific spatial domain I didn't even tell you that but I only chose the middle of the embryo because here you see there's a much more complicated pattern of time in the head the future head of the animal and I cannot explain this pattern at all with the two factors that I have so I have deliberately chosen to focus on only a specific time in space at a space a region in space and also a specific time during development when these two genes interact these are all questions all of them that depend on my purpose what am I trying to do what is my goal here my goal was just to give you a very simple example of a pattern forming process and how you could study that using a systems approach but when you do research you have to wonder what are all the important interactions and factors what am I interested in at what level do I want an explanation and all of these choices not only depend on your system if they also depend on what you want to get out of the system that's very important that's the essence of perspectives and so some people like Dominic Chu have argued that because of all of those choices being dependent on our questions there is no such thing as a system so Dominic is pointing out in a in this wonderful paper which is called Against Systems I'll refer to it in and provide this paper as reading material so so he makes two points one of them is how we draw the boundaries around a system is arbitrary in two ways first of all which factors you include like you see here you could draw the boundary very wide or then focus in on a subset of the system that explains all that you're interested in or you can choose certain subsets of interactions from all the the totality of interactions that you have in a system and how you do this will completely depend on what you're interested in and whether those interactions or factors actually contribute to the phenomenon that you're interested in so Dominic takes this very far here he is and he is saying he's asking the question is there a system and they're sort of in the world out there and his question is no okay there is no such thing as a system out there system definition he writes is a choice that mostly depends on the specific purposes that motivated the modeling existence okay so we may say oh if systems are the main objects of study in biology and they don't even exist what is happening here okay so complexity again just like with Luenton and Levins and Whimsad is seen as the difficulty of making choices about which elements of a system to include in a model the more complex the system the more difficult it is to get your perspective right because the more choices you have to make which is a direct consequence of complex systems having many perspectives that are valid okay so the the father of cybernetics Ross Ashby agrees with Dominic a system is a set of variables selected by an observer it's up to you it's in the eye of the beholder basically so what are we gonna do fuck the system no I want to argue and I'm gonna end this lecture on that that systems are real and I give you a wonderful series of quotes from Austrian systems biologist Paul Weiss which wrap up which are a bit complicated so we'll have to parse them but they pretty much wrap up what a real-world system is Weiss says here pragmatically defined a system is a rather circumscribed complex of relatively bounded phenomena so you you can recognize it sort of it's not easy but you can see that it is something which within those bounds retains a relatively stationary pattern of structure in space or of sequential configuration in time so there is again there's there's a recognizable behavior in nature and you want to understand that behavior and you can bound it to some degree but this happens in spite of a high degree of variability in the details of distribution and interrelations among its constituents constituent units of lower order so the components come and go the interactions come and go but that pattern that overall pattern that you see remains the same behavior at the system level not necessarily one-to-one dependent on behavior at the level of the parts second quote the systems concept is the embodiment of the experience that there are pattern processes I love this quote which owe their typical configuration not to a prearranged stereotype mosaic of single-tracked components of performances not like a machine you don't have predefined components but on the contrary to the fact that the component activities have many degrees of freedom that submit to the ordering restraints exerted upon them by the integral activity of the whole in its pattern systems dynamic so the behavior of the whole is not only not predictable from its components it affects the behavior of the components downwards we'll come back to that when we talk about causation in complex systems the basic characteristic of a system is its essential invariance beyond the much more variant flux and fluctuations of its constituents again it is defined at the systems level not necessarily at the level of the components this is exactly the opposite of a machine yes in which the structure of the product depends crucially on strictly predefined operations of the parts in the system the structure of the whole determines the operation of the parts in the machine the operation of the parts determines the outcome this is why organisms and systems complex systems in general do not behave like a classic mechanical machine okay to wrap up what I wanted to tell you in this lecture is that systems are real they're not just in our heads they are robustly observable pattern processes and that this fits of course perfectly into the definition of reality of trustworthiness of robustness by Bill Wimsa but complex adaptive systems allow for a large number of valid independent perspective so there's not only one way to look at them and we should use a lot of different perspectives to go at them and when formalized in some way those perspectives become models of the system so a formalized perspective of a system on a system is a model of the system and we can use those models to study what the system is capable of doing we'll go into this in the next lecture I hope you'll join me again thanks for listening