BEACON | An NSF Center for the Study of Evolution in Action

BEACON Researchers at Work: What Every Scientist Needs to Know

Posted on December 1, 2014 by Danielle Whittaker

This week’s BEACON Researchers at Work blog post is by University of Texas graduate student Amir Shahmoradi.

Summary: In a world in which science and technological breakthroughs dominate all aspects of almost every individual human life, scientists and researchers are under an ever increasing pressure to cross and expand the borders of human knowledge. As new discoveries require higher levels of precision and reproducibility, excess workload and hyper-competitive work environments have made researchers more prone to human cognitive biases. A solution to this emerging problem is to introduce all graduate students in STEM fields with the limitations of human mind and scientific instruments and their potential role in false positive discoveries and misconduct of scientific research. I suggest that a full-semester course that covers relevant topics including those mandated by NSF as Responsible Conduct of Research should be developed and tailored for each individual STEM field of research and be offered as an integral core course of every graduate program across the world.

Growing up in a traditional and highly religious society, I was drawn from an early age to the romantic mystique of ancient religious and philosophical writings. I joined study sessions and participated in lively discussions with religious scholars. But living in an academic household, I gradually developed a sense of scientific skepticism that led me to question the basic tenets of this knowledge. By contrast, science and mathematics seemed so captivating to me as a teenager for a very simple reason: Science is based on observation, evidence, and mathematics. It is universal, independent of people, society, religion and ideologies.

My passion for science, in particular Astronomy, Physics and Biology kept growing, until I stumbled on a post dubbed “The Same Color Illusion” in Astronomy Picture of the Day (APOD), which profoundly changed the way I view and perceive the world around me ever since. This APOD post showcased a simple example of human cognitive bias and how it can affect our perception of similar and different colors, with a simple clear message: “What human senses perceive of the world, does not necessarily reflect the reality.”

The psychological literature is full of studies that demonstrate how human’s limited senses can result in cognitive flaws and biases in our understanding of the universe. In fact, psychologists have pinpointed many types of biases that affect not only the way we see but how we think about and react to the world around us. Confirmation bias, for example, is the tendency to notice, accept, and remember data that confirms what we already believe, and to ignore, forget, or explain away data that is contradictory to our beliefs. To make things worse, add the (unknown) limitations of instruments by which human probes the universe. The combined effects of human and instrument biases can result in erroneous conclusions and predictions.

Fortunately, many of such biases are now well understood by scientists, in particular, by experimental physicists, biologists and observational astronomers. A worked-out example is the well-known Malmquist bias in observational astronomy. Nevertheless, as our circle of knowledge expands, so does the circumference of darkness surrounding it, bringing new types of instrumental and human cognitive biases with it, that might affect human’s understanding of natural phenomena.

Today, we live in a world that relies heavily on science and technology. As a result, the number of scientists has also grown exponentially rapidly over the past century. With limited funds and resources now available to the community of scientists, the competition and work stress has also increased steadily among researchers.

In such a hyper-competitive atmosphere, scientists are more prone to perception and cognitive biases due to excess workload and stress. There already exist websites, such as Retraction Watch, that regularly report new examples of wrong scientific papers, and papers that contain fake or irreproducible results, forgeries and plagiarism.

The two major funding resources of science in the United States, the National Science Foundation and the National Institute of Health have already stepped in to mitigate the increasing trend that is seen in irreproducibility of scientific discoveries and retractions of scientific articles, before scientists lose the public’s trust in their work. Examples of actions taken include new rules for validating scientific discoveries and mandatory Responsible Conduct of Research (RCR) for all students and postdocs supported by NIH and NSF funds.

Personally, I cannot believe that any scientist in the world would intentionally want to fake results or commit plagiarism or be involved in any other unethical action. Over the past decade, I have witnessed how human cognitive biases can affect the minds and scientific results of numerous scientists. I have seen scientists who insist on the accuracy of their wrong discoveries, and in many cases, I have become convinced that there is no personal intention involved in their stance. I have been very fortunate to work on some specific research projects that opened my mind to many of the limitations that we humans and our scientific instruments face in probing and understanding the universe.

I personally believe the RCR trainings mandated by NIH and NSF can become even more efficient, if they were instead offered as a mandatory comprehensive full-semester course, for all graduate students in all STEM fields, a course that would also cover the myriad of human cognitive biases and instrumental limitations that would meddle with reasoning of every scientist and their understanding of natural phenomena. Regardless of where these students end up, whether academia or industry, whether they are funded by NSF/NIH or not, every student in science programs must learn about the limitations of human mind and its potential adverse effects in scientific reasoning and discoveries.

For more information about Amir’s work, you can contact him at a dot shahmoradi at gmail dot com.

Posted in BEACON Researchers at Work | Tagged BEACON Researchers at Work, cognitive bias, responsible conduct of research | Leave a comment

BEACON Researchers at Work: Engineering life

Posted on November 24, 2014 by Danielle Whittaker

This week’s blog post is by University of Washington graduate student Leandra Brettner.

All living organisms share a universal programming language—DNA. Long strings of unit molecules A’s, T’s, C’s and G’s dictate the unique traits of each individual, but the code is read ubiquitously across each species. This means that a gene that encodes a protein in one organism would encode the same protein if transplanted to another creature. Synthetic biologists use this property to engineer life by doing just that, rearranging genes from different species to program new behaviors into organisms. I am a synthetic biology graduate student in the lab of Professor Eric Klavins, and I work with genetically programmed bacteria, specifically Escherichia coli.

Microbes such as viruses, bacteria and yeast, are cheap and easy to grow, making them excellent platforms for synthesizing traditionally expensive organic chemicals such as fuels, pharmacologicals, and commodities like plastics. By performing the chemistry to create these products in microorganisms, we can potentially both decrease cost and increase sustainability and performance. Researchers like Jay Keasling at UCSF and Angela Belcher at MIT are demonstrating the amazing utility of living chemistry by manufacturing drugs such as artemisinic acid in yeast and building record breaking batteries out of viruses.

However, when we introduce foreign behaviors into cells, we are competing with millions if not billions of years of evolutionary history. Microbes, like all organisms, work hard to maintain the energy balance that supports life. Synthetic programs mess with that equilibrium, limiting the engineering complexity we have currently been able to achieve.

I work on developing ways to increase the complexity of engineered behaviors in microbes by isolating them into working groups—kind of like how factories use assembly lines, everyone has a specific task that contributes to the whole. These division of labor schemes are seen through every hierarchy of biology, from symbiotic bacteria to eusocial insects.

Our system’s goal is to digest complex carbohydrates like those in plant waste and turn it into usable biomass that can go towards producing carbon-based products like the biofuels and therapeutics mentioned, further reducing the cost and making production carbon neutral.

The population of engineered bacteria start out in a consumer state where their only job is to grow and reproduce. Then, every so often, a cell will switch to an altruistic state where it produces an enzyme that breaks down cellulose and lyses to deliver the goods to the extracellular environment. The digested sugars can then be used as food for the consumer cells.

This cooperative architecture has allowed us to build in the complex behavior of novel nutrient use that can be coupled with chemical production in the future.

However, this system suffers from an interesting form of community evolutionary instability called “the tragedy of the commons.” In well mixed culture, any variants that arise that cease to perform the cooperative behavior (cheaters) can still reap the public good provided by the altruists. Because they fail to lyse, the cheaters have an increased fitness advantage and can sweep the population—but to their ultimate demise. Without the altruists, cellulose digestion comes to a halt and the population crashes. Previous work has shown that if, however, there is some spatial organization to the environment, the communal benefit applies only to nearby, closely related cells who are likely fellow altruists. The cheaters are left stranded with limited or no access to the resource. This phenomena, dubbed kin selection, propagates the cooperative behavior through many generations. Members of Professor Ben Kerr’s lab are currently working with my system to investigate if they can evolve strains that exhibit increased cooperation by propagating cell lines in structured environments.

I look forward to continuing to collaborate with the Kerr lab, and potentially extending their research to the design and tuning of new synthetic organisms.

For more information about Leandra’s work, you can contact her at leandra dot brettner at gmail dot com.

Posted in BEACON Researchers at Work | Tagged BEACON Researchers at Work, Biological Evolution, Cooperation, E. coli, genetic engineering, Synthetic biology | Leave a comment

Meet the 2014 BEACON Distinguished Postdocs, Chandra Jack and Will Soto

Posted on November 17, 2014 by Danielle Whittaker

This year, BEACON was fortunate to be able to appoint TWO new Distinguished Postdoctoral Fellows. Meet Chandra Jack and Will Soto!

Chandra Jack

Chandra started working in the Strassmann-Queller lab at Rice University as an undergraduate in the summer after her sophomore year to earn money while volunteering as a member of the Rice EMS. That summer she began research looking at kin discrimination between different genotypes of Dictyostelium purpureum. Originally planning to go to medical school, Joan got her hooked on all things dicty, so she spent a year after graduating as a technician before beginning graduate work in the same lab. She received her PhD in evolutionary biology where her thesis work explored how the population structure of D. discoideum is affected by interactions with related species and with other members of the same species.

Now at MSU, Chandra joined the Friesen’s lab in the plant biology department in August. She was really interested in Maren’s research exploring the mutualism between plants and rhizobia, as well as the many methods being used in the lab. However, she isn’t sure if Maren would have accepted her if she knew her plant reputation included losing one cactus and killing another.

(Currently all of her plants are alive and accounted for.)

Her research will address the role of PMI (Plant-Microbe-Insect) interactions in driving rapid evolution using Medicago polymorpha. She will compare the response of different genotypes of M. polymorpha from its native habitat in Europe to those of invasive genotypes found in North America when they interact with local (N. American) herbivores and rhizobacteria from both environments to look for evidence of genetic variation and to see if genetic variation between the two genotypes results in changes in gene expression. She will also create models that investigate the role of PMI interactions on the relationship between gene expression and plant fitness. Chandra’s work will determine the importance of multitrophic level interactions on rapid evolution and the success of invasive species as they enter new territories.

For more information about Chandra’s work, you can contact her at chandra dot jack at gmail dot com.

Will Soto

Will received his B.S. in biology from California State University, Fresno. Will developed an enthusiasm in microbiology in high school biology classes but became fascinated with evolution as a subject while at CSU, Fresno. Learning about the geological history of the fossil record and the rich biological diversity that evolved through adaptive radiations was exciting to Will. Evolution’s great stories like the “Age of Fishes,” “hopeful monsters,” the Cambrian Explosion, and mass extinctions were intriguing to Will. Additionally, learning about evolution made Will wonder why there were no freshwater echinoderms or freshwater cephalopods, given the tremendous biodiversity of these taxonomic groups. “Evolution has a great folklore and causes one to wonder about the rest of the natural world,” says Will. Will’s interests in microbiology and evolution merged into one. Will was especially interested in prokaryotes due to their colossal genetic and metabolic diversity.

After graduation from CSU, Fresno, Will spent two years in Fred Cohan’s lab at Wesleyan University, where he studied bacterial evolution. “It was in Fred Cohan’s lab that I learned about microbial experimental evolution and developed an interest for the work of Rich Lenski, Al Bennett, and Mike Travisano,” says Will. “When I read the Nature paper by Rainey and Travisano (1998) about adaptive radiation with Pseudomonas fluorescens, I was completely thrilled,” states Will. “Wrinkly spreaders and fuzzy spreaders; here’s another cool story,” he adds. After leaving Wesleyan University, Will entered a PhD program at New Mexico State University in Las Cruces in Michele Nishiguchi’s lab, where he studied the sepiolid squid-Vibrio mutualism. Will pursued a microbial experimental evolution project, where he serially passaged Vibrio fischeri through a novel squid host. “I took a V. fischeri strain indigenous to the Hawaiian bobtail squid (Euprymna scolopes) and serially transferred it through the Australian dumpling squid (Euprymna tasmanica),” claims Will. “I had a great PhD advisor who allowed me complete freedom. I also had a fantastic graduate committee,” says Will. Kathy Hanley, Geof Smith, John Gustafson, and Michele Nishiguchi were all on Will’s dissertation committee. Kathy Hanley is a superb evolutionary biologist, while Geof Smith and John Gustafson (now at Oklahoma State University in Stillwater) are spectacular microbiologists. Michele Nishiguchi provided the expertise on host-microbe interactions, along with the sepiolid squids and bioluminescent V. fischeri.

In 2012, Will became a postdoctoral teaching fellow funded through a grant from the Howard Hughes Medical Institute. He had two excellent postdoc advisors, Mike Travisano and Robin Wright. “I was delighted to be in Mike Travisano’s lab, as he was one of my grad school heroes!” says Will. In Mike’s lab, Will learned the tricks of the trade to microbial experimental evolution. Robin Wright mentored Will in the value of active learning, science education, and how to incorporate research into undergraduate education. “At the University of Minnesota-Twin Cities, I got to teach in an active learning classroom for the first time alongside Robin Wright. The experience was invaluable!” claims Will.

Here at Michigan State University, as BEACON postdoc fellow, Will is working with Chris Waters on developing host infection models between disease-causing marine bacteria (e.g., Vibrio harveyi) and invertebrate hosts (e.g., shrimp) for microbial experimental evolution projects. “Chris and I are trying to take microbial experimental evolution with vibrios to aquaculture,” states Will. Will concludes, “I don’t understand why more microbial experimental evolution work hasn’t been done with the Vibrionaceae. This bacterial family has much to offer in studying evolutionary biology.”

For more information about Will’s work, you can contact him at wsoto at msu dot edu.

Previous recipients of the BEACON Postdoctoral Fellowship are Annat Haber (2013) and Joshua Nahum (2012).

Posted in BEACON Researchers at Work | Tagged BEACON Distinguished Postdoctoral Fellow | Leave a comment

BEACON Researchers at Work: Using evolution to update old software

Posted on November 10, 2014 by Danielle Whittaker

This week’s BEACON Researchers at Work blog post is by MSU graduate student Erik Fredericks. His first BEACON blog post is here.

Engineers in industry are often given the somewhat daunting task of updating legacy software systems. Consider, for a moment, a piece of software used on a daily basis at a particular company, say, Initech. This software was programmed roughly twenty years ago by an engineer that has since left the company, leaving only cryptic comments and mostly unreadable software requirements. However, the importance of this software is paramount to Initech’s success, given its near-ubiquitous use throughout the corporation.

Times have changed over the past 20 years, Initech has gone through some changes, and our software package is quietly reaching the end of its usefulness. However, everybody still continues to use it, even though it is no longer as useful as it once was. Conveniently, a new engineer has been hired and is given his first task: update the software system, thereby extending its effective lifetime. Everybody is already used to it, the TPS reports it generates are spot-on, re-training all employees would be cost-prohibitive, and really, how hard can it be to update software?

After spending weeks gathering new requirements, analyzing the source code, and evaluating the overall state of the framework, our engineer has determined that the easiest way to fix the software system is to use software composition. Composition is a reuse technique for introducing new functionality into existing systems by reusing existing modules from other software systems. For instance, security can be added to a web application by composing in a security module that has been written for another application, where the composition is enabled by transforming the existing data into the format required by the new security module. Moreover, the data being returned must also be transformed into the format required by the legacy application.

Even though the code for the new module has already been written, there still remains many different ways in which to compose the module. For instance, how exactly should the legacy software system send data to the module? How should the data be returned? Is it even written in the same programming language? There are innumerable approaches for transforming data, and our software engineer would rather not spend the extra time and effort figuring out how to accomplish this task. Our engineer decides that an automated approach is best. Enter, context-free grammar-based genetic programming (CFG-GP).

CFG-GP is a search-based technique for automatically generating programs, similar in fashion to the standard genetic program. It comprises a population of candidate solutions that are evolved over time towards a particular goal or set of goals. CFG-GP differs from standard genetic programming in the solution representation. Particularly, each possible solution defined for the genetic program is specified by a context-free grammar. The grammar imposes a structure on generated solutions, effectively reducing the amount of invalid programs typically generated by a normal genetic program and also reducing wasted computing time. For composition, the grammar provides a structure for the many different ways in which data can be transformed for a particular programming language.

As with a typical genetic program, CFG-GP must also specify a fitness function to guide the search process towards an optimal solution. In our case, we leverage software preconditions and postconditions, along with common software engineering metrics, to guide the search process. Preconditions and postconditions define the conditions necessary for a module to be invoked within a program. For example, preconditions can define the data type and order of method parameters, and postconditions can define a method return value or the state of the program following invocation of the method. Together, preconditions and postconditions define what exactly is required for a method to be used by a program, enabling us to use it as part of the fitness function. In our case, candidate solutions must satisfy all preconditions and postconditions.

For example, we may wish to define preconditions and postconditions for a new security module, particularly for encrypting data to be sent to a server. A sample encryption module we wish to compose is defined as follows, in which the encryption function accepts two strings and a string pointer, and returns a boolean value:

<bool> openssl_private_encrypt(String data,
String &encrypted,
String private_key)

We now define the preconditions and postconditions necessary to compose this module. Particularly, the module preconditions represent the format the data is required to be in prior to module invocation. Essentially, we are interested in the parameters accepted by the module:

STRING: data_in
POINTER: data_out
STRING: key

Next, the postconditions for the module must be defined to represent the state of the data following invocation of the module. Here, we are interested in what the module returns back to the legacy software system:

BOOLEAN: return_value

Using these preconditions and postconditions, coupled with further definition of data available and state of the legacy software system prior to and following module invocation, we can successfully compose the security module.

Common software engineering techniques also help to guide the search process at a high level, enabling an engineer to fine-tune the set of optimal solutions. For instance, minimizing the amount of extra function calls, data dependencies, or amount of generated code may be required on systems that have limited memory available. Obfuscation of generated code may also be of interest for high-security systems. Satisfaction of preconditions and postconditions ensures that generated composition strategies are successful, and tailoring of solutions with software engineering techniques can tailor composition strategies for the needs of the engineer.

Thus far, our engineer has created a grammar detailing possible composition strategies, specified the preconditions and postconditions necessary to compose the new module, and executed SAGE to generate a set of optimal composition strategies. It seems that far more work has been incurred by setting up and running SAGE than would have been required to simply hand-code the composition. The benefit, however, is that SAGE can now be run for nearly any composition that the engineer may need to perform in the future. The defined grammar now forms a basis for any future compositions and can be easily extended to other domains that the engineer may encounter. Furthermore, SAGE can also automatically generate valid code in whichever language is required, enabling the engineer to simply copy and paste the generated composition strategy into the legacy framework.

For more information about Erik’s work, you can contact him at freder99 at msu dot edu.

Posted in BEACON Researchers at Work | Tagged BEACON Researchers at Work, Evolutionary Applications, genetic algorithms, genetic programming, software engineering | Leave a comment

BEACON Announces Search for the John R. Koza Endowed Chair in Genetic Programming

Posted on November 4, 2014 by tim

A generous gift to BEACON and MSU by Dr. John R. Koza, the acknowledged founder of the field of genetic programming and author of four books on that subject, has endowed this chair, aimed at recruiting one of the leaders in the field as a new MSU faculty member and BEACON contributor. This endowed chair will help to build BEACON’s strength in evolutionary computation during its second five years as an NSF Science and Technology Center, and help to sustain BEACON after the STC funding is over. BEACON’s other endowed chair, the Koenig Chair, is held by Prof. Kalyanmoy Deb, an acknowledged leader in multiobjective evolutionary optimization, and we hope to fill the Koza Chair in GP with someone similarly distinguished in the field of GP. Together, these chairs will play an important role in strengthening BEACON’s long-term contributions to the field.

Full information about the position is available here.

BEACON’s leaders and members join in thanking Dr. John Koza for this marvelous sustaining gift!

Posted in Job Openings, Member Announcements | Tagged endowed chair, Koenig Chair, Koza chair | Leave a comment

BEACON Researchers at Work: Improving Search Quality in Genetic Programming

Posted on November 3, 2014 by Danielle Whittaker

This week’s BEACON Researcher at Work blog post is by MSU graduate student Armand Burks.

About Me

As a young “army brat,” I was always fascinated with learning how things worked. Trying to figure out how my toys worked, I often found myself amidst a pile of plastic pieces, wires, and motors, desperately trying to perform the magic trick of transforming those pieces back into one while my parents remained none the wiser. To this day, I am still not sure if I ever became more skilled at this trick or at making the toys disappear altogether. Either way, I became a better magician, and my interest in how things worked only grew.

It was this same fascination that led me to study Computer Science, as I could not only safely break things to understand them, but I could also create things from scratch. During my undergraduate studies at Alabama State University, I gained exposure to research through the Historically Black Colleges and Universities Undergraduate Program (HBCU-UP). This program ultimately led me to Michigan State University, where I first learned about Evolutionary Computation. From there, it was “love at first sight.” I was able to make things, break things, and let things evolve, all through the power of research!

My Current Research in Genetic Programming

My current research deals with combating premature convergence in Genetic Programming (GP). Before I delve into premature convergence, a brief summary of Genetic Programming is necessary. GP can be thought of as a special type evolutionary algorithm, where one of the main goals is to automatically create computer programs through evolutionary processes. GP evolves a population of computer programs, where programs are typically executed on an optimization problem and compared based on how close they are to solving that problem.

The major difference between GP and most evolutionary algorithms is way in which the genotype (genetic makeup) of a program is represented in GP. In the most basic form of GP, genotypes are represented by a tree structure because it lends itself well to the creation of programs. As a concrete example, the figure at right shows the genotype of a program that encodes the mathematical expression: x² – 1. When this tree’s genetic code is executed, the function nodes (* and -) operate on the terminal nodes (the input variable x and the constant value 1) to produce an output value. The evolutionary process breeds programs from the evolving population by combining their genetic material in order to discover new programs.

Premature Convergence in Genetic Programming

Premature convergence is one of the biggest problems in Genetic Programming, and evolutionary algorithms in general. To sum it up, premature convergence can be thought of as evolutionary stagnation. Since GP is often applied to optimization problems, one way to gain insight into the difficulty of such a problem for GP is to consider the enormity of the problem’s search space (the large number of possible computer programs that could be generated in attempt to solve the problem). Premature convergence takes place when the population quickly becomes so similar that the algorithm can no longer effectively explore the search space. At this point, it is often very difficult or even impossible for the population to improve in fitness (breed better programs). There are several reasons why this occurs, but the loss of genetic diversity is one of the most notable.

This is a particularly interesting problem in GP because of the tree-based genotype. Previous research has confirmed that not only do traditional GP populations very quickly lose genetic diversity in general, but a large contiguous portion of most of the trees become completely identical in a top-down fashion. This has a very profound and stifling impact on the effectiveness of finding solutions to difficult problems.

Combating Premature Convergence and Improving Search Quality

Fortunately, a lot of research has gone into combating premature convergence, and a lot of great strides have been made towards improving search quality. One example from the evolutionary computation literature is the Age-Layered Population Structure (ALPS). The main idea of ALPS is to use a special concept of age to divide the population into several layers. This allows new genetic material to be effectively inserted into the population and flow upwards from the bottom, while older genetic material evolves and improves as it makes its way towards the topmost layers. This helps to keep the population from becoming too genetically similar while it explores the search space. ALPS and other techniques have been successfully demonstrated in a number of problem domains.

Although there have been a lot of improvements that address premature convergence with varying degrees of success, I believe that there is still a lot of room to do better, and this is my current research focus. It is known that techniques designed to combat premature convergence often incur the cost of taking longer to find the optimal solution to a problem, although these approaches generally increase robustness (the reliability of finding optimal solutions). This is because these approaches avoid prematurely converging by aggressively exploring new areas of the very large search space instead of focusing in on one particular area. With this in mind, one of my research questions is: can we design robust algorithms that are capable of effectively avoiding premature convergence while still finding the optimal solution faster?

For more information about Armand’s work, you can contact him at burksarm at msu dot edu.

Posted in BEACON Researchers at Work | Tagged BEACON Researchers at Work, Evolutionary Applications, genetic algorithms, genetic programming, optimization | Leave a comment

BEACON Researchers at Work: Spatial dynamics of evolution

Posted on October 27, 2014 by Danielle Whittaker

This week’s BEACON Researchers at Work blog post is by MSU graduate student Emily Dolson.

All biological organisms must occupy a single location in physical space. This idea is so obvious that most people don’t give it much thought, but it has important consequences. The spatial location of an organism controls what other organisms it can interact with. The sets of organisms can mate with each other, for instance, has a big impact on the way mutations can spread through a population. Similarly, the sets of organisms that compete with each other is a fundamental component of the selection pressures that they experience. If you consider other types of interactions, such as parasitism, mutualism, and predation, the implications of spatial structure only continue to grow.

As an undergrad, I did a project on the spatial distribution of biodiversity in a forest. We didn’t end up having enough data to draw many conclusions, but I learned a lot about statistical techniques for describing spatial structure. More importantly, I learned that these techniques fascinate me! But perhaps the most important thing that I learned from that project was that there were benefits to collaborations between computer scientists and ecologists.

I was doing this project at the end of my sophomore year, the time when students at my college had to declare a major. And it just so happened that at this time, for the first time in many years, I was unsure what field I wanted to go into. Since high school, I had been sure that I wanted to be a quantitative ecologist. But during the first two years of college, I had taken some computer science classes and fallen in love with them. Double-majoring happened to fit fairly neatly into my schedule, but I worried about how I was going to choose one field over the other when it came time to apply to grad school. Through the collaborations that it encouraged, the spatial biodiversity project allayed these fears. It gave me an opportunity to apply my skills from both fields and get advice from professors in both departments. Suddenly, I not only felt that I didn’t have to choose between fields, I felt that I could do more for science as a whole if I wasn’t forced to.

Fast-forward a few years and now I’m a second year PhD student in the Ofria lab, pursuing a dual degree in Computer Science & Engineering and Ecology, Evolutionary Biology, & Behavior. I’m working on a number of projects, but the ones that are perhaps nearest and dearest to me are surprisingly similar to my undergrad project; they, too, deal with spatial structure and diversity. The biggest difference is that now I’m studying these phenomena in the context of evolution. I like studying the process of evolution in as general a case as possible, because it means I can do work that simultaneously has implications for both evolutionary computation and biological evolution.

The other big difference between my work now and my work as an undergrad is that now I do most of my research in computational systems. The system I most commonly use is the Avida Digital Evolution Platform. Basically, Avida consists of a grid of virtual CPUs running self-replicating sequences of assembly code (Avidians). Any new organisms generated over the maximum population size overwrite pre-existing organisms, resulting in pressure to replicate quickly. Because the instruction that Avidians use to copy their genomes has the potential to introduce mutations, evolution occurs. Of course, Avidians do not actually exist in physical space. For the sake of answering questions about spatial dynamics, however, we choose to treat the grid as if it represents their spatial proximity to each other. This is a neat and intuitive way of achieving a gradient of interactions between organisms. Having such a gradient is useful both because it has theoretically interesting properties and because these properties are also present in natural systems as a result of them existing in physical space.

This image shows an environment containing 16 evenly spaced circular patches of 8 different resources. Cooler colors represent locations in the grid where a greater number of resources are rewarded. As you can see, this one environment contains an intricate variety of spatial niches.

So far, the primary spatial project that I’ve been working on is one that I started for the BEACON spring course last semester, with Samuel Perez and Audra Chaput. Ecologists have long debated the reason that we see so many different species in ecosystems that don’t seem to have very many different niches (i.e. potential roles for a species within in an ecosystem). It turns out that there are a lot of reasons that this occurs, but one that is particularly commonly suggested and yet uncommonly tested is the idea that spatial heterogeneity allows an ecosystem to support more species. Spatial heterogeneity is the idea that most ecosystems have areas within them that are different from each other. For instance, while an ecosystem as a whole might have two resources present in relatively equal amounts, there might be pockets within that ecosystem in which one of these resources is much more common than the other. This has the potential to create many additional niches. Since even quantifying spatial heterogeneity in nature is challenging, exploring its implications in Avida makes a lot of sense. In Avida, we are able to set up patches of different resources and overlap them in different ways, creating a wide variety of niches (see image above). We can then explore the way that diversity changes over time across this spatial environment, in comparison with more or less heterogeneous environments. Because spatial strategies could be used to maintain a diversity of solutions in evolutionary computation, this work has practical implications for ecology, evolutionary biology, and evolutionary computation.

A number of other BEACONites are also doing interesting work on spatial evolutionary dynamics, and I am fascinated to see where it leads.

For more information about Emily’s work, you can contact her at dolsonem at msu dot edu.

Posted in BEACON Researchers at Work | Tagged Avida, BEACON Researchers at Work, biodiversity, Biological Evolution, Digital Evolution, ecosystems, spatial relationships | Leave a comment

BEACON Researchers at Work: Evolution of plasmid host range

Posted on October 20, 2014 by Danielle Whittaker

This week’s BEACON Researchers at Work blog post is by University of Idaho postdoc Wesley Loftie-Eaton.

I stumbled into the world of plasmids at my alma mater, the University of Stellenbosch in South Africa. My advisor, Prof. D. E. Rawlings, asked me to determine the biological implications of a very specific mutation within the origin of replication of two isogenic plasmids (Loftie-Eaton and Rawlings, 2009; Loftie-Eaton and Rawlings 2010). What started as a very simple and specific question quickly enthralled me with its true complexity. During that time I learned that you can’t understand plasmids, or anything in biology for that matter, without the context of evolution; and what better place to study plasmid evolution than at the University of Idaho alongside Dr. Eva M. Top and under the umbrellas of IBEST and BEACON.

Plasmids are genetic elements that replicate separately from the bacterial chromosome. Many can transfer horizontally between diverse bacteria, often resulting in the spread of antibiotic resistance (red insert on plasmid) to pathogens (red bacterium).

In case you’re not a plasmid geek like me, plasmids are generally circular (linear forms also exist) DNA entities that replicate autonomously in bacteria. The smallest plasmids are but a few hundred base pairs while the largest are over a million base pairs long (Kado et al., 1998). Based on sequence information, about half of plasmids are either self-transferable by means of conjugation or can be transferred (mobilized) by conjugative plasmids, while the remainder appear to be non-transferable (Smillie et al., 2010). Many plasmids are cryptic or do not encode any non-essential genes while others may carry genes that provide benefit to the bacterial host, allowing it to occupy and proliferate in an otherwise hostile niche (Kado et al., 1998). Depending on whether a plasmid provides its host with benefit or cost, they can be likened to symbionts or parasites, respectively, and like symbionts and parasites, they have a host range that is either narrow or broad.

Projects I am currently working on are focused on a) elucidating the molecular mechanisms that evolve to permit plasmids to shift, contract or expand their host range and b) to understand how broad host range plasmids proliferate and spread in complex microbial communities. The latter will not be discussed here. Besides the fundamental nature of these questions, they are relevant to human health due to the rampant plasmid-mediated spread of antibiotic resistance amongst pathogens.

When a plasmid conjugates to a naïve host it may not necessarily persist unless there is selection for its maintenance in that host. Failure to persist could be due to inefficient replication, poor partitioning or segregational loss of the plasmid during cell division, or plasmid-containing cells can simply be outcompeted by plasmid-free cells if the plasmid imposes a cost on the host. We know, however, that plasmid and host can rapidly adapt to each other while under strong selection for plasmid maintenance, after which the plasmid can continue to persist for prolonged periods even when the strong selection is removed (Bouma and Lenski, 1988; Dahlberg and Chao, 2003; Sota et al., 2010). Panels A and B in the figure below demonstrate exactly this; an initially unstable plasmid-host relationship evolved a more stable phenotype in less than 200 generations and after ~600 generations the plasmid was highly stable in that host¹. Not clear from this figure (due to me not showing the full assay data) was that in the absence of antibiotic selection the ancestral plasmid tended towards extinction, however, a persistent relationship in which the plasmid was maintained in ~10% of the population due to horizontal transfer evolved within the first 100 generations (panel C, below)! Together with collaborators Sam Hunter and Haruo Suzuki I am currently working on elucidating the genetic changes behind this increase in persistence and collaborator Jose Ponciano is working on means to quantify how easy or difficult it is to switch between trajectories of persistence and extinction.

While under antibiotic selection for plasmid maintenance, plasmid stability evolved rapidly and resulted in a persistent state wherein the plasmid was maintained even when the antibiotic selection was removed. (A) Plasmid stability measured over 10 days in the absence of antibiotic for evolving cultures sampled every 100 generations [G] over the course of the evolution experiment. (B) Summary of the endpoint [day 10] data for each stability assay for three replicate lineages. (C) A prediction of whether the plasmid will persist or go to extinction in the absence of antibiotic selection for its maintenance.

In another experiment (different plasmid and host)² we found that a transposon encoded on a plasmid native to that host ‘jumped’ to the introduced plasmid, which was being maintained under antibiotic selection. The result was increased stability of the introduced plasmid, even in the absence of antibiotic selection, and loss of the native plasmid. Encoded on the transposon are a toxin-antitoxin (TA) system and a resolvase, both of which we have now shown to promote plasmid stability (Loftie-Eaton et al., in prep.). Even more significant, however, was that the evolved plasmid that acquired the transposon was completely stable in other beta- and gamma-proteobacteria in which its ancestor was unstable. Thus, acquisition of a transposon encoding stability functions resulted in an apparent expansion of the plasmid’s host range and more broadly, we demonstrated the interplay and fate of genes that could arguably be labeled as “selfish”.

In yet another system one of the outcomes was a deletion mutation within the plasmid’s origin of replication. Though much work remains to be done, this mutation has me extremely excited. The type and location of the mutation is the same as the mutation I set out to understand during my PhD. The exciting part is that the plasmids I studied then belong to a different family and evolved in the environment, whereas here the mutations occurred during experimental evolution in the lab. Preliminary results showed that this deletion abolished the plasmid’s ability to replicate in a previously permissive host, which already is novel information, however, if my hypothesis based on my previous research withhold scrutiny, then these results will demonstrate that what we observe during experimental evolution in the lab also occurs in nature, and vice versa. Validation!

Though far from complete, what we’ve learned thus far is that plasmid host-range can evolve quite rapidly, that such rapid changes tend to occur through mutations that result in gain or loss of function and that there are multiple molecular solutions that can lead to stable plasmid-host relationshi

ps. From the perspective of a plasmid geek this is fascinating, but from a medical perspective it’s concerning; once a multidrug resistance plasmid has established in a population it intends to stay, even if the antibiotics that initially selected for its maintenance are removed from the system! However, by accumulating more data of this kind we hope to define general patterns in the evolution of plasmid host range that can aid in the development of novel drugs to inhibit the spread and establishment of multidrug resistance plasmids.

Acknowledgements:

¹The experimental work was done by undergraduate Kelsie Bashford.

²Much of the experimental work was done by undergraduates Ryan Simmons and Stephen Burley as well as our lab manager Linda Rogers.

References

Loftie-Eaton, W., and D. E. Rawlings. 2009. Comparative biology of two natural variants of the IncQ-2 family plasmids, pRAS3.1 and pRAS3.2. J Bacteriol 191:6436-6446.

Loftie-Eaton, W., and D. E. Rawlings. 2010. Evolutionary competitiveness of two natural variants of the IncQ-like plasmids, pRAS3.1 and pRAS3.2. J Bacteriol 192:6182-6190.

Kado, C. I. 1998. Origin and evolution of plasmids. Antonie Van Leeuwenhoek 73:117-126.

Smillie, C., M. P. Garcillan-Barcia, M. V. Francia, E. P. Rocha, and F. de la Cruz. 2010. Mobility of plasmids. Microbiol Mol Biol Rev 74:434-452.

Stewart, F. M., and B. R. Levin. 1977. The population biology of bacterial plasmids: a priori conditions for the existence of conjugationally transmitted factors. Genetics 87:209-228.

Ponciano, J. M., L. De Gelder, E. M. Top, and P. Joyce. 2007. The population biology of bacterial plasmids: a hidden Markov model approach. Genetics 176:957-968.

Bergstrom, C. T., M. Lipsitch, and B. R. Levin. 2000. Natural selection, infectious transfer and the existence conditions for bacterial plasmids. Genetics 155:1505-1519.

Bouma, J. E., and R. E. Lenski. 1988. Evolution of a bacteria/plasmid association. Nature 335:351-352.

Sota, M., H. Yano, H. M, Julie, G. W. Daughdrill, Z. Abdo, L. J. Forney, and E. M. Top. 2010. Shifts in the host range of a promiscuous plasmid through parallel evolution of its replication initiation protein. ISME J 4:1568-1580.

Dahlberg, C., and L. Chao. 2003. Amelioration of the cost of conjugative plasmid carriage in Eschericha coli K12. Genetics 165:1641-1649.

Loftie-Eaton, W., Burleigh, S., Simmons, R., Rogers, L., Hunter, S., Settles, M., Ponciano, J.M. and Eva Top. Transposition of a toxin-antitoxin system and plasmid-host coevolution increase plasmid persistence and host range. In preparation.

For more information about Wesley’s work, you can contact him at wesleyl at uidaho dot edu.

Posted in BEACON Researchers at Work | Tagged antibiotic resistance, BEACON Researchers at Work, Biological Evolution, drug resistance, genetics, host-parasite coevolution, plasmid | Leave a comment

Why lactation rooms matter

Posted on October 14, 2014 by Danielle Whittaker

This post is by UW graduate student Carrie Glenney, and is cross-posted from the UW biology graduate student blog Science Positive.

A lack of access to lactation rooms might be a widespread issue for women in academia. As demonstrated in the Biology Department at University of Washington, it can also be a relatively simple problem to solve. Ensuring lactation room access for all women in academia would send a very important message: we support you.

Although women make up more than 50% of science PhDs earned, they are more likely than men to leave the academic sciences at every stage on the way to obtaining a tenured position at a college or university. A study by Marc Goulden, Karie Frasch, and Mary Ann Mason suggests that becoming a mom may be one major determinant of this trend: married mothers with young children are less likely than both single and married women without young children, and 35% less likely than married fathers, to achieve tenure at a college or university.

PhD students and post-docs specifically may face obstacles that make having a child feel antithetical to continuing in academia: a lack of paid and limited unpaid parental leave, limited affordable childcare, or lack of advisor or departmental support. Although many of these issues affect fathers as well as mothers, surveys with University of California post-doctoral scholars at both the beginning and end of their programs showed that of those who had children during their post-doc, women were twice as likely as men to change their career goal away from “professor with a research emphasis”.

Why does having a baby seem to hinder a woman’s academic career but not a man’s? What unique obstacles do female PhD students and post-docs with children face on the pathway to a career in academia?

I had my son during the third year of my PhD and I anticipated and experienced some of the issues addressed above. But the complete lack of one resource in particular surprised me the most. In many cases, this resource is actually a legal right. It’s relatively straightforward to provide. And I think mandating its presence in every department would go a long way towards making mothers in academia feel more accepted and supported.

Lactation rooms.

Returning to work after having or adopting a baby can be HARD, no matter where you work, and no matter how much you might love your job. You might be operating on a handful of hours of sleep a night. Your body might be still healing from childbirth. You might feel a mix of emotions about returning to work and physically being apart from your baby. You might not feel physically or emotionally ready to come back to work but at the same time feel that you don’t have a choice. And for those mothers who are able and chose to breastfeed, there’s the where, when, and how to navigate around pumping.

Logistically, continuing to breastfeed even when you’re not physically around your baby can be tricky. For those who aren’t familiar, here are some of the logistics. A woman’s body produces only the amount of milk it thinks the baby needs. The body forecasts how much milk it thinks it should make based on how much was used in the past. If less is used, less is made in the future. In order to convince your body to continue making the same amount of milk even when your baby isn’t around, you need to remove milk at least as frequently as your baby would (in the range of every 2 to 4 hours) and sometimes more often (because pumps don’t remove milk as efficiently as a baby). Pumping frequently and for enough time is important for maintaining milk production, but it’s also essential to prevent a very painful infection called mastitis. Once the milk is pumped, most moms store it in the fridge or freezer and it’s given to their baby through a bottle when the mom isn’t around to breastfeed. Removing the milk requires a pump (most women use electric pumps), plastic pieces that need to be washed after each use, cold storage, and a stress-free environment.

The last requirement may surprise you. Milk flow actually requires the release of the hormone oxytocin (the “love” hormone), a process usually stimulated by baby. In the absence of baby, women use other methods to encourage milk flow (like looking at a photo of her baby), but stress, fatigue, or even being cold can make this very difficult. Therefore, a private, clean, comfortable space with a locking door is essential. Ideally, there is also a sink for washing parts, a fridge for milk storage, and desk space so work can continue if desired. Most PhD students and post-docs share an office with others so they cannot or understandably might not want to pump in their office. Often women are relegated to bathrooms to pump, but any place where you would not want to eat lunch is also not a place where a mom will want to pump milk for her baby. Pumping can take 10-30 minutes a session and has to happen every 2-4 hours, so location convenience is also an important factor for accessibility.

Source: milkitkit.com

When Hannah Kinmonth-Schultz (also a PhD student and mom) and I decided to ask the Biology Department for a lactation room, we first conducted a departmental survey to determine the need. We found that the need was far greater than we’d anticipated: 7 of the 86 total female respondents (8%) said they anticipated a need for a lactation room within the next 12 months or have a current need. 81% of respondents who already have children said they would have benefited from a departmental lactation room. Convenience was an important factor: only 1 mom reported using the campus lactation facility located about a five minute walk away. Other moms used the bathroom (and one mom reported throwing all her milk away for sanitation purposes) or borrowed other people’s offices if they didn’t have their own. One mom commented, “I can’t tell you how many uncomfortable places I have pumped!”

Beyond the need for a space to meet the physical needs of pumping moms, providing
an accessible lactation room for every mom in academia sends an important message: we support you and your family. Many women in academia who want to have children are hearing a different message: this is not the time or place for a baby. In addition to the implicit message sent by the lack of resources and support from the institution itself, many women may be discouraged outright by mentors, advisors, or even peers. I have had personal experiences with this* and I’m sure others have as well. Academia is an environment where there’s often little distinction between “work” and “life”, let alone a balance, and it’s easy to feel like family planning should take the back burner or you’ll risk harming your career. So returning to work to find that the most basic of necessities, like a place to pump, aren’t available to you can make you feel like your choice to have a child is not supported. In addition, some advisors may not understand the importance of providing the time and physical space for pumping (although my advisor was wonderfully accommodating); or a female graduate student might feel understandably reticent about broaching the topic of breastfeeding with a male advisor. In the comment section of our departmental survey, one mom addressed the emotional toll of having no place to pump by saying “having a lactation room in the biology department while I was nursing would have gone a long way towards helping me feel accepted. I felt extreme guilt and very alone. As a result I think that my research progress suffered much more than it would have if impacted just by my new time constraints.” Another said “stopping pumping because of lack of convenient facilities was especially hard for me and not ideal for baby.” Even though breastfeeding is hard, many moms want to keep at it because it can have such wonderful benefits: money savings from not needing to buy formula (~$100/month), nutritional and immunity benefits for baby, and the baby-mother bonding that breastfeeding facilitates. Yet ¼ of the respondents in our survey said they quit breastfeeding earlier than they wanted to because they had nowhere to pump.

This is just a snapshot demonstrating the need for a lactation facility in one department at one university, but it wouldn’t be a stretch to imagine that other departments have a similar scenario- and they may not even know it.

As Stanford University stated in 2006 “… a woman’s prime childbearing years are the same years she is likely to be in graduate school, doing postdoctoral training, and establishing herself in a career.” If we want to support women in academia, one of the most important things we can do is acknowledge that some will want to start their families during this timeand to support them when they do. I’m proud to be of a department that is taking steps in this direction by establishing a lactation room** (it should be available soon). Even when space is limited or money is tight, a little creativity can help convert a space to be used for this purpose. For example, our department is currently converting the departmental break room in the Kincaid building into a shared space that can function as a private lactation room during non-meal hours.

Of course, we need a multi-pronged approach to support women in academia and to support both women AND men who want to start families while in academia. Ensuring access to lactation rooms is only one step… but a meaningful one.

Other than the departmental survey we conducted last year, I don’t know of any studies that have been done to look at whether there is a widespread lack of access to lactation rooms in universities and colleges. In hopes of getting the dialogue about this issue rolling: if you are (or know of) someone*** at a university or college who needs a lactation room and doesn’t have access to one, or has a story about starting a lactation room at another department or institution, or just wants to lend your support, please visit our facebook group Lactademia.

* I would like to extend a thank you to my own advisor (Ben Kerr) and the Kerr lab. They have given me nothing but support, including moving offices around so I would have a place to pump. Thank you!

** Special thanks to all who helped make the UW Biology Department lactation room a reality:Hannah Kinmonth-Schultz, Diversity committee members Horacio de la Iglesia, Andrea Prado, Rose Ann Cattolico, Greg Wilson, Linda Martin-Morris, Sarah Eddy, Julian Avila, and Camilla Crifo; Executive committee members Toby Bradshaw, Michele Conrad, Carl Bergstrom, David Perkel, and Jennifer Ruesink; Graduate Program Director Marissa Heringer.

*** The issue of access to lactation rooms also affects lab technicians, faculty, and many others in the academic world. This article focuses primarily on post-docs and graduate students because those are the individuals I’ve interacted with the most about this issue and the levels of academia with which I am most familiar. But the ultimate goal is “access for all.”

Posted in BEACON Researchers at Work | Tagged diversity, lactation, women in academia | Leave a comment

BEACON Researchers at Work: The Evolution of Gene Distribution in Bacteria

Posted on October 13, 2014 by Danielle Whittaker

This week’s BEACON Researchers at Work blog post is by MSU graduate student Nathan Ward.

One of the strengths of BEACON is its focus on interdisciplinary research. Through gathering together scientists and educators from different backgrounds, BEACON can address problems that people from one field alone cannot solve. One example of this is Avida, which combines insights from computer science and biology to simulate evolution. Avida is an example of what can be achieved when computer scientists and biologists collaborate together; problems that seem difficult in one discipline are much easier to solve when approached with insight from another discipline.

Like Avida, the work I am doing right now is interdisciplinary. I am currently a PhD student in the computer science department with a background in computer engineering, however, one of my PhD advisors, Dr. Julius Jackson, is a professor in the Department of Microbiology and Molecular Genetics at Michigan State University. The work I am trying to do right now is to create software that will simulate the evolution of bacteria starting roughly four billion years ago. The major goal of our software is to discover how evolution shapes the distribution and organization of genes on the bacterial chromosome. Because of this, we named our software: Gene Distribution Lab.

Evolution selects for the organisms with the fittest characteristics for their given environmental niche to survive. Likewise, evolution selects for the chromosomes that have the best organization of genes. One feature of chromosomal organization that evolution appears to select for is genes that are functionally related to be next to each other spatially on the chromosome. The observation that bacteria often have functionally related genes close to each other on the chromosome is called the gene clustering phenomenon.

Although there are many different models that scientists have proposed to explain the gene clustering phenomenon, we are starting our investigation by looking at one model specifically. The lab I work for has proposed a model called the Limited Protein Mobility Model. This model states that when genes of related function are close together, the proteins they code for will also be close together, furthermore when these proteins are close together they can process the reaction they both interact in better than if they are far apart.

Although my background prepared me to write the code for Gene Distribution Lab, grasping the underlying biology was a challenging but necessary step at the start of this project. With my background in engineering and computer science I often find it easier to grasp problems of biology through thinking about these problems in terms I am familiar with. So to really understand the Limited Protein Mobility Model, I wanted to put it in context of an engineering problem.

Metabolic pathways in bacteria function a lot like assembly lines. First of all, precursors in the cell are initially processed by the pathway; this can be thought of as the “raw material” that is modified by the assembly line. At each step in the assembly line, a worker will perform some operation on the unfinished product, such as soldering on wires, reshaping the material, adding on paint, or something of that nature. Only after the last step will the final product be useful for what it is intended for.

Likewise, in each step in a metabolic pathway, a protein will modify the unfinished product, that is, the intermediate, such that the intermediate is closer to its form as the desired end product. Proteins will operate on the intermediates, which is analogous to how workers in a factory operate on the unfinished product to bring it closer to completion.

One big difference between a metabolic pathway and an assembly line is that pathways operate in a different way due to the scale difference and the fluid medium that they are in. Intermediates and proteins are suspended in the cytoplasm of the cell, and move around randomly due to the laws of chemical diffusion. So imagine, instead of a belt carrying everything around the assembly line, all the workers and the unfinished products were drifting around randomly in a pool of thick gel. It would be very hard to complete the final product this way. Each worker could only modify the unfinished product if the right pieces drifted over to them, and even then they couldn’t send the product directly over to the next worker in line. The problem would be exacerbated if the workers were all very far apart, because it would take longer to initially start creating end products and many pieces of the unfinished product (the intermediates), would simply drift away out of reach of any worker. We propose a similar process promotes the gene clustering phenomenon. Genes that produce proteins in the same pathway will tend to cluster together on the chromosome because it is easier to make end products when proteins are close together.

Gene Distribution Lab tests this prediction by simulating the evolution of early bacteria. These bacteria start off with a very small chromosome and a number of genes they are required to have to code for proteins in a pathway. The chromosomes then undergo segment duplication, segment deletion, and point mutations which have the potential to disable genes they have. An organism dies if it loses the last copy of a gene necessary for survival. So far we have verified that organisms in Gene Distribution Lab are able to accumulate and dissipate clusters in their chromosome, and soon we will be able to show the effects of the Limited Protein Mobility Model on the selection pressure to maintain clusters.

For more information about Nate’s work, you can contact him at wardnath at msu dot edu.

Posted in BEACON Researchers at Work | Tagged BEACON Researchers at Work, Biological Evolution, chromosomes, Digital Evolution, gene clustering, proteins | Leave a comment