John L Jerz Website II Copyright (c) 2015

Robustness and Evolvability in Living Systems (Wagner, 2005, 2007)

Andreas Wagner

"insight into the evolution of robustness in biological systems can assist in evolving devices that are robust to a defined spectrum of faults."

"Invasion resistance is the notion of robustness I am concerned with here... it is... the invasion of nonresident species - that drives these communities to a state of higher invasion resistance."

"One important commonality between robustness in living and other systems is obvious: Something - either the state of a system or the system itself - persists in the face of perturbations... The robustness of nonliving systems must have a different origin, an origin that can be subsumed in one word: self-organization."

"evolutionary searches... usually do not start from scratch, but tinker with systems that already serve some other purpose."

JLJ - Only a geneticist will present pictures of the heads of flies (p.168 - eeeeuw...) to prove a point. Wagner writes before becoming sidetracked by his genetic networks trainwreck-of-a-concept, and before innovating his innovability concept. I think he strayed too far from the adaptive capacity and evolutionary experimentalism concepts discussed by other authors. If you pop the top on any concept of innovation and gaze inside, IMHO, you will discover simply 'tricks that work', and nothing more. For the case of an innovation-producing system, we see merely, 'tricks' (variation and selection in its many forms), that produce 'tricks that work'. Or put my ideas aside and you can follow Wagner into his neutral networks concept -- see how he believes the biological world manufactures innovations.

Wagner argues like Darwin and biologists do - they cite examples from nature. A concept is neither true nor false when supported this way - it is merely worthy of consideration, in a field dedicated to produce understanding of complicated subject matter.

IMHO, the traditional variation and selection concept is a powerful enough explanation for evolution in any form it is presented - living or non-living systems included. Every new creature is itself an innovation. All organisms struggle to survive. Every so often, one arrives on the scene that is so built that it - for whatever reason - leaves more offspring, which in turn produce more offspring. This is a tough trick to perform, considering the competition that exists. Is this itself an "innovation"? I would say that it is more of an unexpected match-to-environment, we are merely calling it an innovation. The ice fish is also an innovation but more of a freak fish that just happened to survive in a very cold, harsh environment with limited competition for food. Selection over the millions of years - really just differential-offspring-producing - has favored those innovations that just-happen-to-be/are-better-matched to their environment.

Wagner's foundational neutral space concept (see quotes from Schuster paper http://fontana.med.harvard.edu/www/Documents/WF/Papers/sequences2shapes.pdf below, should really be read to make sense of his work), and tracks well with my "multiple good moves" concept for game theory. Having multiple ways to proceed in a complex game of strategy almost guarantees that one of these ways - for reasons initially unknown and unknowable ahead of time - will develop unexpectedly favorable results down stream, and will eventually be "selected" by your future self as part of a plan to "go on". A young rabbit develops two survival abilities - the freezing posture, and the running away tactic. It does not know which one to use ahead of time, but presses one button or the other when confronted by a predator, and eventually survives. A trapped rabbit in a cage will alternate between these behaviors (running away turns into darting back and forth) because it does not know what else to do - other than chewing on the cage with its teeth. Survival in a complex game of strategy involves making general purpose threats and exploring tactical opportunities for both sides. Survival, recovery from the unexpected - and possibly victory if the opponent makes a series of small or large mistakes - are the possible results of such behavior.

Wagner's 'invasion resistance' is a great concept which can be used in the theory of complex games of strategy - when in doubt of what to do, increase your own invasion resistance, and work to erode that of your opponent, by examining multiple ways the adaptive capacity to mobilize coercion can unexpectedly change. You may not find any quick routes to success, but you will be more or less 'ready' if or when one is uncovered.

p.6 Most problems the living have solved have an astronomical number of equivalent solutions, which can be thought of as existing in a vast neutral space... A neutral space is a collection of equivalent solutions to the same biological problem. Such solutions are embodied in biological systems that ensure an organism's survival and reproduction.

p.7 Robustness and neutral mutations are key to evolutionary innovation... Robust biological systems permit many neutral mutations, mutations that do not affect a specific system function. However, these mutations can affect other properties of the system, properties that may be the source of future detriment or benefit, and also the source of evolutionary innovations.

p.8 Many natural systems below and beyond living organisms show great robustness to changes in their parts. Such robustness can also increase over time, but the cause is usually self-organization instead of natural selection... Many of the mechanistic principles that underlie robustness in living systems can also be observed in man-made, engineered systems

p.8 It has been said that nothing is ever new. [JLJ - This is the first time I have heard this...]

p.10 if you are a nonspecialist interested in the questions I pose, this book may be for you.

p.45 blind evolutionary searches in sequence space will not be effective at finding rare structures in its vast expanses.

p.45 Two RNA sequences whose nucleotides differ at k positions are also called k-mutant or k-point neighbors of each other, because they can arise from each other by k changes of single nucleotides. A sequence that folds into a frequent structure typically has many one- and two-mutant neighbors that fold into the same structure... Such neighbors are called neutral neighbors. As a consequence, one can hop via single nucleotide mutations from sequence to sequence, without ever leaving the set of sequences folding into the same minimum-free energy structure. Put differently, many sequence pairs in this set can be connected through a series of neutral mutations, mutations that leave secondary structure unchanged. This observation prompted Schuster and collaborators to call this set a neutral network (491).

p.88 Robustness is only one of several features of a biological system that affect its ability to evolve... Another such feature is modularity... Modularity occurs on all levels of biological organization.

p.93 two main approaches can provide information on a biological system's robustness. The first consists of many experimental perturbations of the system's parts... The second consists of comparing systems in related species, systems that derive from a common ancestor and that represent different solutions to the same biological problem.

p.160 Robustness can increase in evolution, if the network's gene expression pattern is under the influence of selection favoring the persistence of this pattern.

p.162 The studies I review first take a circuitous route to detect robustness. Specifically, they show that development is robust to ubiquitous genetic variation, but not by detecting robustness under normal conditions. No, they generate conditions... under which such robustness breaks down. By doing so, they also show that robustness can be... controlled, which makes its change in biological evolution possible.

p.175 This chapter... makes two central observations. First, there are many and sometimes radically different ways to build the same body or body part... The second observation is that such variation is a key to innovation... Most biological problems have a vast number of different solutions, and many of these solutions harbor the seeds of innovation to solve other problems.

p.195 I argue here that the following concept of a neutral space can provide a unified view of robustness in many different systems I discussed earlier:

A neutral space is a collection of equivalent solutions to the same biological problem. It can also be thought of as a set of alternate configurations of a biological system, configurations that solve the same problem.

p.195 robustness of biological systems... is rooted in the structure of neutral spaces.

p.203-204 The higher one climbs in the hierarchy of organismal organization, the less information about the robustness of biological systems is provided by systematic experimental and computational perturbations. Instead, one has to rely increasingly on a limited proxy for perturbations, namely evolutionary comparisons of systems that perform the same function in different ways. While not strictly proving robustness of any one system, such comparisons can suggest that there are multiple different ways of solving the same problem, and thus a large neutral space associated with the problem.

p.214 natural selection cannot resolve infinitely small fitness differences.

p.215 evolutionary searches... usually do not start from scratch, but tinker with systems that already serve some other purpose.

p.216 In sum, I posit that a biological system's robustness (or lack thereof) is explicable by these two factors: whether evolution found a frequent or rare solution to a problem, and whether adaptive evolution toward increased robustness occurred within the neutral space associated with this solution.

p.217 The word "evolvability" has two main usages... According to the first of them,

a biological system is evolvable if its properties show heritable genetic variation, and if natural selection can thus change these properties.

A second usage ties evolvability to evolutionary innovations:

a biological system is evolvable if it can acquire novel functions through genetic change, functions that help the organism survive and reproduce.

p.218-219 Neutral genetic change... is commonly understood as genetic change that does not affect an organism's fitness... how can we determine whether a mutation does not affect fitness? ...fitness is difficult to define properly, and nearly impossible to measure rigorously

p.220 The last paragraphs show that neither fitness nor a biological system's performance - that is, all aspects of it - can be measured in practice.

[JLJ - Maybe, but it can be estimated by a diagnostic test - a useful concept for game theory. An elementary school conducts a fire drill to estimate the fitness of the organization to respond to an unexpected fire. The results are not perfect, but point to areas in need of improvement. A bank conducts a stress test in the event of a financial crisis, etc.]

p.252 To see what all this has to do with the evolution of robustness, consider that natural selection (of anything) needs genetic variation within a population. Thus, natural selection of robustness needs variation in robustness. Put differently, only in populations polymorphic for robustness can selection change robustness. A variety of studies arrive at this insight... from different angles

p.268-269 The above two models regard two maximally different mechanistic causes of robustness: distributed robustness and redundancy of parts... These differences suggest that there may be no fundamental theory of how robustness evolves, if such a theory is required to take into account the different architectures of biological systems. The reason is that the mechanistic cause of robustness strongly influences how robustness can evolve. However... Simple yet general principles can still apply to all biological systems.

p.297 One important commonality between robustness in living and other systems is obvious: Something - either the state of a system or the system itself - persists in the face of perturbations... The robustness of nonliving systems must have a different origin, an origin that can be subsumed in one word: self-organization.

p.300 robustness in communities develops in ways fundamentally different from how robustness evolves in populations of organisms under natural selection.

p.300 Community ecology is concerned with the factors that determine which species occur in a community and how abundant its members are. As usual, when studying robustness of a community to perturbations - in ecology often called resilience (71, 72) - it is important to have specific community features and perturbations in mind. Here, I will be concerned with robustness of the coarse community structure - the identity of species in a community. The perturbation at issue is the invasion of a community by a new species. This is clearly only one of many perturbations one could study, but it will serve to make the central point.

p.304 Invasion resistance is the notion of robustness I am concerned with here... A community's robustness or invasion resistance can be defined as the proportion of (regional) species that are not already in the community and that cannot invade the community. A community is completely invasion resistant if no regional species can invade it... Increased invasion resistance occurs as a statistical trend in assembling communities

p.305 How can we understand the increase of invasion resistance...? ...Clearly, it is... the invasion of nonresident species - that drives these communities to a state of higher invasion resistance.

p.307 communities can eventually attain complete invasion resistance. No species in the region can invade them.

p.308 Despite these substantial knowledge gaps, the example of community assembly serves to illustrate one central point. Perturbations in various forms can push a system toward increased robustness to the very same perturbations it encounters.

p.309 if the world is rife with robust systems, it is because fragile systems are fleeting.

p.312 inelegance may be necessary to render a biological system robust... while it may be possible to design engineered systems simply and elegantly, robust engineered systems may be no simpler than biological systems.

p.315 the example also suggests that the absence of completely rational, premeditated system design may favor the origin of distributed robustness.

p.315 Evolutionary principles can be used to develop devices that perform a specific computation (279). The same principles can also render a device robust to failure in its parts.

p.319 Both distributed robustness and its evolution can play as important a role in engineered systems as in biological systems... insight into the evolution of robustness in biological systems can assist in evolving devices that are robust to a defined spectrum of faults. The simplest such insight is that of a minimum necessary rate of either random or targeted "mutations."

[JLJ - Ok, for game theory these 'mutations' become instead intelligently-generated "promising or what if" explorations coupled with an experienced sensing of conditions which merit more or less time for diagnostic testing. We aim to test our posture - or the posture of an opponent - to reconfigure and produce 'invasion resistance' under a wide variety of conditions. Sports teams call this a 'scrimmage' - but it must be intelligently constructed and interpreted.]

From Sequences to Shapes and Back: A Case Study in RNA Secondary Structures, Peter Schuster, Walter Fontana, Peter F. Stadler, Ivo L. Hofacker

p.279 Folding can thus be viewed as a map between two metric spaces of combinatorial complexity, a sequence space and a shape space.

p.282 A neutral path is defined by a series of nearest neighbour sequences that fold into identical structures.

p.283 The consequences of our results for natural and artificial selection are immediate. We predict that there is no need to systematically search huge portions of the sequence space. In the particular example of RNA molecules of chain length 100 the characteristic ball contains some 10²⁷ sequences, which is only a fraction of 10^-33 of the entire sequence space. Almost all structures are within reach of a few mutations from a compatible sequence (average: 7.2), and even in reasonable proximity of any non-compatible random sequence (≈18). The conclusion, thus, is that optimization of structures by evolutionary trial and error strategies is much simpler than is often assumed. It provides further support to the idea of widespread applicability of molecular evolution [citations...]. The existence of networks of neutral paths percolating the entire sequence space has strong implications for (molecular) evolution in nature, as well as in the laboratory. Populations replicating with sufficiently high error rates will readily spread along these networks and can reach more distant regions in sequence space.