3rd Annual Big Data in Biology Summer School

Posted on March 9, 2016 by hageyt

This post is by UT Austin graduate student Rayna Harris

The Center for Computational Biology and Bioinformatics at The University of Texas at Austin is proud to host the 3rd Annual Big Data in Biology Summer School May 23–26, 2016.

SummerSchool-2016_Legal

The 2016 Summer School offers eleven intensive courses that span general programming, high throughput DNA and RNA sequencing analysis, proteomics, and computational modeling. These courses provide a unique hands-on opportunity to acquire valuable skills directly from experts in the field. Each course will meet for three hours a day for four days (either in the morning or in the afternoon) for a total of twelve hours.

UT Austin and BEACON students, faculty and staff receive a great discount off the regular fee!

This year a number of BEACONites are participating as instructors, TAs, or organizers. They include: Laurie Alvarez, Dhivya Arasappan, Daniel Deatherage, Emily Dolson, Nicole Elmer, Benjamin Goetz, Rayna Harris, Arend Hintze, Hans Hofmann, Sean Leonard, Kasie Raymann, and Stephanie Spielman. We would like to acknowledge BEACON for supporting the Computational Modeling to Study Evolution in Action course taught by Arend and Emily.

Click here for more information and to register

Great introductory courses:

  • Introduction to Core Next Generation Sequencing (NGS) Tools
  • Introduction to Proteomics
  • Introduction to Python
  • Introduction to RNA-seq

Bioinformatic courses:

  • Bash Beyond Basics
  • Genome Variant Analysis
  • Machine Learning Methods for Gene Expression Profiling Analysis
  • Medical Genomics
  • Metagenomic Analysis of Microbial Communities

Computational Modeling:

  • Computational Modeling to Study Evolution in Action
  • Protein Modeling Using Rosetta

New in 2016:

  • Bash Beyond Basics: This course will focus on being more productive in the Bash shell. We will learn about regular expressions, Unix utilities like cut/sort/join, awk, advanced piping, process substitution, string manipulation, and Bash scripting. Learn to love the command line and increase your productivity with rapid manipulation of bioinformatic data!
  • Metagenomic Analysis of Microbial Communities: This course surveys the Python software ecosystem and familiarizes participants with cutting-edge data science tools. Topics include interactive computing basics; data preprocessing and cleaning; exploratory data analysis and visualization; and machine learning and predictive modeling.
  • Clinical Genomics: This course will introduce a selection of genomics methodologies in a clinical and medical context. We will cover genomics data processing and interpretation, quantitative genetics, association between variants and clinical outcomes, cancer genomics, and the ethics/regulatory considerations of developing medical genomics tools for clinicians. The course will have an optional lab component where participants will have the opportunity to explore datasets and learn basic genomics and clinical data analysis.
  • Computational Modeling to Study Evolution in Action: This class is about the study of evolution using computational model systems. We will use two different systems for digital evolution: Avida and “Markov Gate Networks” exploring many different possibilities of using computational systems for evolution research. Participants will gain a hands-on introduction to the Avida Digital Evolution Research Platform, a popular artificial life system for biological research and the Markov Gate Network modeling framework to study questions pertaining to neuro-evolution, behavior, and artificial intelligence.

Click here for more information and to register

Posted in BEACON Researchers at Work, BEACONites, Education, Member Announcements | Tagged , , , , , | Leave a comment

Introducing BEACON’s New Science Outreach and Communication Postdocs

Posted on March 8, 2016 by hageyt

This post is by MSU postdoc Wendy Smythe.

Wendy Smythe and Minnie Kadake deploying a CTD sensor in Southeast, Alaska.

Dr. Wendy F. Smythe is an environmental scientist who came to BEACON from CMOP who looks at how microbes influence their environment, by examining geochemistry, microbial ecology, microbial diversity, and biomineralization of iron and manganese oxidizing microorganisms. She also directs a Geoscience Education Program within her tribal community located in Southeast Alaska.

One of Wendy’s roles at BEACON will focus on outreach and diversity in STEM disciplines, with specific interest in recruitment and retention of Native American/Alaska Native into STEM. Wendy is Kaigani Haida from the Village of Hydaburg, Alaska and has worked for eight years to couple Traditional Ecological Knowledge with STEM in a culturally competent way to monitor the health of local rivers and coastal ecosystems. (For more info. Visit: http://sustainablesoutheast.net/category/communities/hydaburg/ and http://www.stccmop.org/education/k12/geoscience).

Wendy will also be continuing her previous research from Yellowstone National Park and Southeast, Alaska, by expanding her research focus to evolutionary biology of microorganisms from different research sites and analyzing metagenomes from two extreme metal rich groundwater ecosystems. This research blends evolution, microbiology, ecology, geochemistry, and education working with Dr. Judi Brown Clarke and Dr. Ashley Shade as her mentors.

Wendy and Minnie sampling from a manganese depositing hot-spring in Yellowstone National Park. Manganese specimen collected from hot-spring and scanning electron microscopy of the surface of the rock showing manganese minerals, microbial cells, and extrapolysaccharide.

Posted in BEACON Researchers at Work | Tagged , , , , | Leave a comment

2nd Active LENS Workshop: June 2016

Posted on February 29, 2016 by hageyt

This post is by MSU postdoc Mike Wiser 

avida-ED-logoThe 2nd annual Avida-ED Active LENS Workshop will be held at Michigan State University June 9-11, 2016 in East Lansing, MI. The purpose of this workshop is to train instructors in the use of the Avida-ED software package, developed to help students learn about evolution and the nature of science, so that workshop participants can both implement classroom interventions using this software and also train other educators. Teams of two will learn to use Avida-ED and how to best incorporate it into courses that they teach. Travel and expenses related to the workshop will be covered for the 20 workshop participants as part of an NSF-funded IUSE grant.

Avida is a digital evolution software platform used to study evolutionary processes, and harness evolution to solve engineering problems. Avida-ED is a free, user-friendly version of Avida developed specifically for educational purposes, with a graphical user interface and visualizations that allow the user to observe evolution in action. (See http://avida-ed.msu.edu/ for more information and to download a copy of the software.) Organisms within this software (Avidians) are self-replicating computer programs, competing for computational resources supplied by the environment. Their replication is imperfect, resulting in mutations in some of their offspring, which may alter the ability of those organisms to make use of their environmental resources. Populations studied over the course of generations therefore display all of the elements necessary for evolution by natural selection: variation, inheritance, selection, and time. Avida-ED thus provides not a simulation of evolution, but an actual instance of it.

Avida-ED has been developed for undergraduates and advanced placement high school students to learn about the nature of science and evolution in particular. Users have significant control of the environment, and are able to change parameters such as the world size, the mutation rate, and what resources are available. Individual organisms can be saved in a virtual freezer, analyzed individually to watch how they perform tasks and replicate themselves, and used to start new evolutionary runs. Because digital organisms grow and divide much faster than even the fastest microbes, Avida-ED allows users to test evolutionary hypotheses over the course of hours or minutes. By generating hypotheses, collecting data, and analyzing results, users gain experience not just with concepts in evolution, but with the nature and practice of science as a whole.

Workshop participants will join a growing community of educators using digital evolution to let their students directly observe evolutionary processes through inquiry-based exercises that advance reform-oriented active learning. Participants will develop new lesson plans and will help collect assessment data from their classroom implementations. They will help disseminate materials and train other science educators; financial support is available for this. At least one member of each pair will attend a 1-day follow up meeting at MSU in early summer 2017 to report on their experience.

The team application form for the Active LENS Workshop must be completed online on the following page: http://avida-ed.beacon-center.org/. Applications should be submitted no later than March 7, 2016. If you have any questions or difficulties with the application, contact Michael Wiser (mwiser@msu.edu).

Posted in BEACONites, Education, Member Announcements | Tagged , , , | Leave a comment

An Evolutionary Computation Perspective at AAAI 2016

Posted on February 26, 2016 by hageyt

This post is by UT Austin grad student Elliot Meyerson 

aaai_posterI spent February 12-17 in Phoenix at the 30th AAAI Conference on Artificial Intelligence. AAAI covers artificial intelligence “broadly conceived”, with focus on “traditional topics such as search, machine learning, planning, knowledge representation, reasoning, natural language processing, robotics and perception, and multiagent systems.” Evolutionary Computation (EC), a large subfield of AI, tends to be underrepresented at mainstream AI conferences like AAAI; rather EC is featured at more specialized venues such as GECCO, PPSN, and CEC. As an EC researcher, attending AAAI gave me a chance to update my contextualization of EC in the broader field, and find inspiration in places I don’t normally look; a few things stood out in particular.

General evolutionary computation framework

EC covers modeling and optimization techniques inspired by natural evolution. Generally, EC systems maintain a population of individuals encoded as possible solutions to a computational problem or task. Individuals are selected for recombination and mutation based on their fitness, a value corresponding to the success of the individual in the task. Intuitively, evolution will naturally progress to produce individuals that maximize this fitness. In this form, fitness is the only information collected from the task, which is otherwise considered a black-box.

A variety of black-box optimization techniques aside from EC populate the AI community. At AAAI, I found only two papers that explicitly used EC, including the one I presented [1,2]. Other authors were interested in theoretical results that could help bridge the gap between EC and the more theory-oriented AI community. Their papers considered derivative-free algorithms, that use operations analogous to mutation, but forsake crossover, enabling formal proofs of their probabilistic approximation abilities (under certain assumptions about the black-box) [3,4].

Example of individual generation steps in Bayesian optimization

Another black-box approach with nice theoretical properties and even more prevalent at AAAI was Bayesian optimization (BO). Given all the individuals evaluated so far, instead of generating new individuals through mutation and crossover, BO always generates the individual that maximizes some probabilistic acquisition function, often the expected improvement in the task. Under certain assumptions, BO is theoretically guaranteed to converge to the true global optimum, a quality many EC researchers only dream of. Also, in EC, maintaining and understanding diversity is an important goal; some BO and other techniques at AAAI also made use of the popular formalization of submodularity to prove the approximate efficacy of a diverse collection of solutions. BO was showcased in an invited talk by Andreas Krause, in which he demonstrated its applicability to robotics and protein design, in which BO outperformed previously tried EC approaches [5]. It was encouraging to see the success of such automation in design for synthetic biology, a domain I am interested in pursuing.

Information produced inside of a black-box function can be harnessed to improve optimization.

However, I believe synthetic biology domains are among many for which optimization can be qualitatively advanced by automation techniques that go beyond the black-box, incorporating additional information about the task, the behavior of individuals in the task, or features of the evolutionary process. As an example, novelty search is an approach to escape local optima by rewarding individuals that demonstrate novelty with respect to what they do on their way to receiving their fitness score. Along with providing a powerful search methodology, insights gained in development and analysis of novelty search can then help inform our understanding of natural evolution, e.g., with respect to the evolution of cognition [6] and the utility of extinction [7].

General algorithm configuration framework

At AAAI, beyond-black-box techniques arose in the domain of algorithm configuration (AC), particularly in an excellent tutorial led by Frank Hutter. AC is the problem of automatically optimizing the settings of an algorithm for a particular application. In algorithms across optimization and machine learning, the choice of settings can have huge effects on the algorithm’s behavior and performance. AC works by testing out various settings when applying an algorithm to a set of task instances, and using the results to select settings that will hopefully perform well on future tasks. For EC researchers, AC can be a very useful tool, as EC algorithms often have several parameters, whose optimization can provide both improved performance and insight into how each evolutionary mechanism practically affects evolution across a variety of tasks. The most popular AC techniques use BO [8], and there are competitive EC approaches as well [9]. Beyond the black-box, some techniques used extensive analysis of task features to predict which settings will be optimal for novel classes of future tasks. Hutter also suggested using learning curve information to preemptively terminate unpromising experiments. Like protein design, algorithm configuration consists of fixing an initial state, and then watching a complex hard-to-predict temporal process unfold, which leads to the fitness output. There is a ton of additional information produced by this temporal process that can be potentially harnessed to improve search.

I was very inspired by BO and AC, and have already started messing around with one BO-based AC tool, Spearmint, and working towards using it in my research :). Due to differences in technical details and language, it can often be difficult to take advantage of related research in disparate fields. All in all, an immersive week in the world of AAAI proved to be an effective recontexualization opportunity, aka shake-up. I’m sure there’s such a thing as too many shake-ups, but I think most of us probably don’t get enough of them. Also, in Phoenix the hiking was good.

Some other unrelated but fun stuff from AAAI for researchers:
Semantic Scholar. From Allen Institute on AI. Goal: make lit-search more efficient
Auto-poster-generation [10]: I will never manually make a poster again (note: the poster presenting this paper was not auto-generated).

AAAI 2016 poster on automatically generating posters.

References:
[1] Braylan, A.; Hollenbeck, M.; Meyerson, E.; Miikkulainen, R. Reuse of Neural Modules for General Video Game Playing. AAAI 2016.
[2] Singla, A.; Tschiatschek, S.; Krause, A. Noisy Submodular Maximization via Adaptive Sampling with Applications to Crowdsourced Image Collection Summarization. AAAI 2016.
[3] Yu, Y.; Qian, H.; Hu, Y. Derivative-Free Optimization via Classification. AAAI 2016.
[4] Qian, H.; Yu, Y. Scaling Simultaneous Optimistic Optimization for High-Dimensional Non-Convex Functions with Low Effective Dimensions. AAAI 2016.
[5] Romero, P. A.; Krause, A.; Arnold, F. H. Navigating the protein fitness landscape with Gaussian processes. PNAS 2013.
[6] Lehman, J.; Miikkulainen, R. Overcoming Deception in Evolution of Cognitive Behaviors. GECCO 2014.
[7] Lehman, J.; Miikkulainen, R. Extinction Events Can Accelerate Evolution. PLOS ONE 2015.
[8] Snoek, J.; Larochelle, H.; Adams, R. P. Practical bayesian optimization of machine learning algorithms. NIPS 2012.
[9] Ansótequi, C.; Sellman, M.; Tierney, K. A gender-based genetic algorithm for the automatic configuration of algorithms. CP 2009.
[10] Qiang, Y.; Fu, Y.; Guo, Y.; Zhou, Z.; Sigal, L. Learning to Generate Posters of Scientific Papers. AAAI 2016.

Posted in BEACON Researchers at Work | Tagged , , , , | Leave a comment

Manipulating evolution to conserve species

Posted on February 22, 2016 by hageyt

Sarah Fitzpatrick fishing for guppies in Trinidad.

This post is by MSU Postdoc Sarah Fitzpatrick working at the Kellogg Biological Station

Consider a native fish population in a small headwater stream with low genetic diversity due to genetic drift and founder effect (loss of variation that occurs when a new population is established by a small number of individuals). High levels of inbreeding occur because most individuals in this population are close relatives. A dam is put in downstream, erasing the chance of occasional natural connectivity and gene flow from a different population. Drainage from an upstream agricultural plot slowly leaks a toxic pesticide into the stream. Imagine your job is to keep this natural resource—the native fish population—alive and healthy. What’s your plan?

I study evolution through the lens of conservation (and vice versa). How can we manipulate our understanding of ‘evolution in action’ in order to buy time for imperiled populations and facilitate adaptation during rapid environmental change?

Dry intermittent reaches like what is pictured here in Big Sandy Creek prevent connectivity among fish populations.

Humans and climate change have modified natural connectivity patterns of many organisms, producing dire consequences for some. When populations become isolated they tend to be more vulnerable to risks associated with small population size. Combined with further habitat loss and environmental disturbance, this perfect storm of risk factors can culminate in dramatic population decline. We are faced with many situations similar to the scenario I described above. Ninety-three percent of vertebrate species listed under the U.S. Endangered Species Act exist in fragments from a formerly connected range.

I’ve mostly thought about this question in the context of freshwater fish populations that live in headwater streams. Headwater environments make fascinating natural laboratories for studying evolution in the wild because 1) multiple replicate populations often exist across neighboring streams; 2) headwater-restricted populations tend to be isolated from the homogenizing effects of gene flow and have high potential for local adaptation; but 3) they are often colonized by one or several individuals, experience strong genetic drift, and may need occasional gene flow to adapt and persist.

From a conservation perspective, headwater environments are incredibly important yet fragile ecosystems that harbor high biodiversity and are threatened by pollution, physical barriers like dams, and drying from groundwater removal or drought.

Both of the fish species I will highlight from my research occur in small headwater populations, yet they tell quite different stories. My goal is to link the evolutionary biology we’ve learned about one to the conservation biology and management plan of the other. The first is the Trinidadian guppy, an invasive species on six continents, but also a textbook example for studying rapid evolution in the wild (Magurran 2005). The second is the Arkansas darter, a Great Plains endemic in steep decline, and a candidate for listing under the Endangered Species Act.

In Trinidad we took advantage of experimental guppy translocation experiments (Travis et al. 2014) to test the effects of gene flow from divergent populations on locally adapted traits, fitness, and population dynamics in native downstream populations. First, we found evidence for high and rapid gene flow downstream from all six historical introduction sites, yet guppies maintained locally important trait differences that we know to be adaptive based on extensive previous work in the system (Fitzpatrick et al. 2015). Then, we focused on two native populations from headwater tributaries that were downstream from the most recent translocations. We began studying these populations before the introductions and gene flow took place and found that these native populations had tiny effective population sizes (Ne = 2-10) and were likely experiencing inbreeding (Fitzpatrick et al. 2016).

Male guppies from the mark-recapture monitoring project

Gene flow began as non-native guppies swam or were washed downstream and began mating with the native guppies. Our field team visited the two focal populations each month for two and a half years and caught all guppies over 14 mm (about the width of your thumbnail) using traps, mask and snorkel, and butterfly nets. All fish each month were weighed and photographed and all new recruits to the population were given a unique colored tattoo under a microscope and had three scales removed for genetic analyses before being returned to their exact site of capture. In total, over 10,000 guppies from the two streams were individually marked, monitored throughout their lifetimes, and could be classified using molecular markers as a pure native guppy, a pure immigrant, or a hybrid. This study was novel in its ability to capture the initial and long-term effects of gene flow on survival and population dynamics in replicated populations in the wild.

Thick black lines indicate total number of guppies > 14mm captured in each stream over time. Grey boxes correspond to the timeframe in which every individual was genotyped at microsatellite loci for classification into genetic ancestry groups. Colors show the number of individuals in each genetic group caught each month.

Despite gene flow from guppy populations that were originally divergent and adapted to a different environment, genetic rescue (an increase in fitness caused by the introduction of new genetic variation) was documented in both streams. Monthly population sizes skyrocketed from under one hundred to over one thousand individuals, genetic diversity increased substantially, and importantly, much of the success could be attributed to hybrid guppies ( Fitzpatrick et al. 2016).

The types of experiments and monitoring we were able to accomplish in Trinidad are simply not an option for a threatened species like the Arkansas darter. However, the small, native guppy populations we studied could be thought of as proxies for other genetically isolated, imperiled populations. For example, we know that Arkansas darters exist in small, genetically isolated populations and that drought and groundwater removal are causing populations to become even more fragmented (Fitzpatrick et al. 2014). Given the increasing consensus that genetic rescue works, under the right conditions (Whiteley et al. 2015; Frankham 2015; Fitzpatrick et al. 2016), I argue that a management plan involving assisted gene flow is worth serious consideration.

But, many unknowns remain. Deciphering the grey area of when gene flow is needed to increase fitness and provide a demographic boost versus when it results in homogenization, or worse, reduces fitness through introducing maladaptive alleles, is a major challenge. Model systems for studying ‘evolution in action’, like the Trinidadian guppy, might become increasingly crucial for conservation if understanding and manipulating evolutionary processes indeed proves to be one way to curb unprecedented rates of biodiversity loss.

Fitzpatrick, S. W., Gerberich, J. C., Angeloni, L. M., Bailey, L. L., Broder, E. D., Torres-Dowdall, J., Handelsman, C. A., López-Sepulcre, A., Reznick, D. N., Ghalambor, C. K. and W.C. Funk. (2016), Gene flow from an adaptively divergent source causes rescue through genetic and demographic factors in two wild populations of Trinidadian guppies. Evolutionary Applications. doi: 10.1111/eva.12356
Fitzpatrick SW, Crockett H, Funk WC (2014) Water availability strongly impacts population genetic patterns of an imperiled Great Plains endemic fish. Conservation Genetics, 15, 771–788.
Fitzpatrick SW, Gerberich JC, Kronenberger JA, Angeloni LM, Funk WC (2015) Locally adapted traits maintained in the face of high gene flow. Ecology Letters, 18, 37–47.
Frankham R (2015) Genetic rescue of small inbred populations: meta-analysis reveals large and consistent benefits of gene flow. Molecular Ecology, 24, 2610–2618.
Magurran AE (2005) Evolutionary ecology: the Trinidadian guppy. Oxford University Press, Oxford.
Travis J, Reznick D, Bassar RD et al. (2014) Do Eco-Evo Feedbacks Help Us Understand Nature? Answers From Studies of the Trinidadian Guppy. Advances in Ecological Research, 50, 1–40.
Whiteley AR, Fitzpatrick SW, Funk WC, Tallmon DA (2015) Genetic rescue to the rescue. Trends in Ecology & Evolution, 30, 42–49.

Posted in BEACON Researchers at Work | Tagged , , , , , , | Leave a comment

Evolutionary signal processing and its application for brain information flow modeling

Posted on February 13, 2016 by hageyt

This post is by MSU PhD Candidate Jinyao Yan.

Jinyao Yan, an MSU BEACON PhD student working in the Statistical Signal Processing Lab, ECE department, but soon to depart to continue her Ph.D. research at the Janelia Research Campus, HHMI.

There are two aspects of biomedical engineering which attract me: the quest for new knowledge, and its power for fostering innovation and promoting society’s development. These pursuits hold my heart; and I am fortunate to follow both passions. Biomedical signal processing appeals to me because it has a mathematical foundation while being transformational in nature. It fuses together my fascination for science and my strong desire to positively influence people’s lives.

During the past half-century, signal processing (SP) has contributed tremendously to 21-st century medical science and practice. Medicine without techniques like magnetic resonance imaging, computed tomography, and ultrasound is almost unimaginable. A rich set of theories and methods are based on linear, time-invariant (LTI) models, and it is these LTI models that have largely supported the spectacular technological change we have witnessed over a few short decades. However, the 21-st century SP engineer is increasingly likely to encounter system analysis and design problems in which LTI models are insufficient. For instance, biomedical systems are complex and are generally nonlinear and time-invariant [1, 2]. The modeling of biomedical systems therefore presents significant challenges not overcome by classical linear methods. In recent decades, intricate research has begun to produce methods for analyzing and modeling isolated classes of nonlinear systems. Biologically-motivated solutions are one extremely compelling current example of this trend. However, this vast class of models still presents many challenges, especially in their application to living systems.

For the last three years, together with my PhD advisors Prof. Erik Goodman and Prof. John Deller, I have worked on evolutionary biomedical signal processing. Our research is concerned with the development of new methods for nonlinear system identification, with broad applicability to modeling problems in biomedicine. The research was initially motivated by the demand for new modeling and evolutionary signal processing strategies for the detection of disease signatures in very low signal-to-noise ratio data [3]. We developed a novel theory and method for nonlinear system identification from measurements under large and complex noise conditions. The approach integrates three modeling and identification strategies: linear-time-invariant-in-parameters (LTIiP) models, set-based parameter identification, and evolutionary algorithms for optimization [4, 5].

Specifically, we treat models as chromosomes: A LTIiP model is the phenotype of a chromosome, a binary sequence in which each bit indicates the presence or absence of a particular gene. Each gene codes for a particular regressor function in the model. A viable model is one with parameter values that allow it to effectively produce the observed output from the observed input. The parameters which appear in the phenotype represent regulators of gene expression, the desired expression being the linear mix of regressors that give the model the highest survival potential. As in nature, survival depends on the inherent suitability of an individual’s genetic makeup to meet the challenges of the environment (reflected in observations), and also in the realization of that genetic potential through an effective parameter set.

The parameter sets result from the set-membership processing of the data. The set-membership algorithms provide sets of feasible parameter vectors rather than a single point estimate. These sets restrict the parameters to those that are possible in light of observations, and they determine the range and statistical viability of the chromosomes.

Following these concepts, the LTIiP model identification problem is formulated into an evolutionary algorithm framework – in particular, a genetic algorithm. The algorithm starts with a random population of chromosomes. Based on the genetic makeup of each chromosome, the feasible set of parameters is deduced using set-membership estimation. Measurable set properties are then used to assign fitness values to each chromosome, and the fitness value is used in the selection process.

Unlike conventional model identification focused on the estimation of parameters, this framework simultaneously addresses selection of the model structure and the parameter estimation. Moreover, a very significant advantage of the algorithm is the lack of need for assumptions about stationarity or distributional characteristics of noise. The ability to identify the correct model with unbiased parameters under complex noise conditions makes the algorithm transformational for practical biomedical data analysis.

Figure 1: System identification results: True data (continuous curve) and estimated data (dash-dot curve). The identified model shows great tracking ability.

We tested the algorithm on several simulated systems and practical datasets, and it exhibits great tracking performance (Fig. 1). We are now applying the algorithm for identifying nonlinear, effective brain connectivity. Advances in neuroimaging and electrophysiological recording have produced a wealth of image and signal data from different brain regions. One goal is to identify sets of brain regions that are simultaneously involved in the processing of a task. Given that the brain transmission is fundamentally nonlinear at the level of individual cell dynamics [6], exploring if information is encoded in highly-nonlinear ways in the brain is essential.

The dataset we are currently working on consists of electroencephalogram (EEG) data of cognitive control signals (Fig. 2). Much research has shown that our brain responds differently when we make a mistake, and this response is called Error Response Negativity [7]. When a mistake is made, the ACC part of the brain, whose role is seen as monitoring and detecting problems, will send a cognitive control signal to the dlPFC to assign more attention and implement adjustments to address problems. Thus, there is a directional and temporal relationship between the ACC and dlPFC. Some contemporary theories also suggest the cognitive control is nonlinear [8]. The objective is to identify this hypothesized brain information flow using our developed framework.

Figure 2: Nonlinear Causal Effective Connectivity Models of the Cognitive Control Networks of the Brain. Left: DLPFC, ACC and related areas of human brain and its cognitive roles (picture credited to https://brmlab.cz/project/brain_hacking/tdcs/pfc). Center: EEG 64 scalp sensors. Upper right: correct response; Lower right: error response.

Besides this application, we are also looking for other applications for neuronal system modeling. Thanks to the great support of my advisers, I am honored to receive a Janelia Graduate Research Fellowship, from the Janelia Research Campus near Washington, D.C. Janelia is part of the Howard Hughes Medical Institute and includes almost 500 scientists pursuing research in fundamental neuroscience and imaging (https://ece.msu.edu/news/jinyao-yan-received-janelia-graduate-research-fellowship). I believe this great opportunity will bring new ideas and possibilities to apply our evolutionary identification methods for biomedical signal processing.

 

[1] E. R. Dougherty. Translational science: epistemology and the investigative process. Currentgenomics, 10(2):102, 2009.

[2] D. T. Westwick and R. E. Kearney. Identification of nonlinear physiological systems, volume7. John Wiley&Sons, 2003.

[3] B. D. Fleet, J Yan, J. R. Deller Jr., D. Knoester, M Yao, and E. D. Goodman. Breast cancer detection using haralick features of image reconstruction from clinical data of ultra-wideband microwave signals. In Proc. 3rd Work shop on Clinical Image-based Procedures: Translational Research in Medical Imaging (CLIP), 2014.

[4] J. Yan, J. R. Deller Jr., M. Yao, and E. D. Goodman. Evolutionary model selection for identification of nonlinear parametric systems. In Proc. 2014 IEEE China Summit and International Conf. Signal and Information Processing, pages693–697, 2014.

[5] J. Yan and J. R. Deller, Jr., NARMAX Model Identification Using a Set-theoretic Evolutionary Approach, Signal Processing, An International Journal of, Elsevier, 2015.

[6] Hodgkin, A. L., & Huxley, A. F. (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. The Journal of physiology, 117(4), 500.

[7] J. S. Moser, H. S. Shroder, C. Heeter, T. P. Moran, and Y. Lee. Mind your errors evidence for a neural mechanism linking growth mind-set to adaptive posterior adjustments. Psychological Sciene, page 0956797611419520, 2011.

[8] Y. Liu and S. Aviyente. Quantification of effective connectivity in the brain using a measure of directed information. Computational and Mathematical Methods in Medicine, 2012.

Posted in BEACON Researchers at Work | Tagged , , , | Leave a comment

Studying the Evolution of Division of Labor with Digital Organisms

Posted on February 6, 2016 by hageyt

This post is by MSU Postdoc Heather Goldsby.

Heather Goldsby, an MSU BEACON postdoctoral researcher working in the Devolab, Kerr, and Hintze labs.

Why do you have different types of cells in your body? Why do honeybees perform different roles, including forager, undertaker, nurse, and queen? Why do factory workers perform jobs as specific as putting on one part for a car? Why do we engineer robots to do different tasks in support of the same mission? All of these are examples of division of labor where individuals take on specific roles and cooperate to survive and thrive. I first became fascinated by division of labor when I realized it underpins all of our economy – and most of our daily interactions at work. My interest only grew as I began to notice it everywhere. Originally, my research started by using division of labor as a tool in algorithms I developed. Then, I started to work on creating evolutionary algorithms that employed division of labor to solve a problem.

As I continued my studies, I learned that division of labor is a key component of major transitions in evolution, and thus of great interest in biology. Major transitions in evolution occur when formerly distinct lower-level entities become linked (either by staying together or by forging bonds) and reproduce and compete as one higher-level entity [1]. For example, major transitions include single cells transitioning into multicellular organisms, solitary insects transitioning into eusocial colonies, and the formation of the eukaryotic cell. For some of these transitions, in particular the ones where genetically similar lower-level entities stay together, a big challenge is how to evolve to specialize and take advantage of the benefits of division of labor.

Because of its importance to science and engineering, I want to better understand how and why division of labor evolves. To do this, I use the Avida digital evolution platform [2]. Its rapid generation times and experimental control enable me to place digital organism (analogous to cells or ants) into groups, apply different evolutionary pressures, and observe how and when division of labor evolves. Using this approach, I’ve studied several aspects of the evolution of division of labor, including: (1) The evolution of temporal polyethism [3]: a form of division of labor used by some bees, where individuals change the task they performed based on their age; (2) The evolution of reproductive division of labor [4]. In particular, we became interested in why do somatic (body) cells evolve when they are evolutionary dead ends? (3) What is the role of task-switching costs in promoting specialization [5]? I’m going to go into greater detail on this question to illustrate how we use digital evolution to study division of labor.

Do task-switching costs promote the evolution of division of labor? Task-switching costs are penalties (in terms of time or resources) associated with changing from performing one type of task to another. For us, they might be the amount of time it takes to shift from checking email or Facebook, to get back to writing a paper. For example, for a bee within a colony, they could include the amount of time it takes to travel from one place in the hive to another, the morphological overhead in changing roles (building new glands, etc.), or even cognitive overhead. To study this question, we placed individual digital organisms into a colony, where the colony as a whole was required to perform a variety of tasks to successfully compete with other colonies. We ran treatments that varied the task-switching costs. We observed that treatments with low task-switching costs evolved generalist organisms: an organism performed many types of tasks. In contrast, when higher task-switching costs were applied, specialist organisms that evolved only one type of task evolved. In an unexpected twist, the specialist individuals also evolved task-partitioning behavior where one individual passed on the results of a task to another individual who used the solution as a building block to perform a more complex task. We see this behavior in both human assembly lines and also insect colonies, such as leaf cutting ants.

Fig. 1. A task-partitioning system evolved by a digital colony. Digital organisms (squares) perform tasks and send messages (solid lines), including task results. In this case, the organisms evolved to send task results to neighbors, who, in turn, used the information to perform more complex tasks.

What fascinated me about the task-partitioning behavior evolved by the individuals within colonies was that while the colony as a whole could perform seven different types of tasks, when placed alone, the individuals could only perform one type of task (Fig. 1). The loss of functionality at the lower-level individual and the emergence of functionality at the level of the colony indicates that task-switching costs could favor both the evolution of division of labor and also a shift in autonomy from a lower-level to a higher-level unit. This shift in autonomy is a fundamental component of major transitions in evolution.

This project highlights how digital evolution can contribute to studies of division of labor and major transitions in evolution. Now, working with colleagues, I’m expanding this research to both understand other evolutionary pressures that favor division of labor and also to see a major transition in evolution unfold in real time.

References:

  1. J. Maynard-Smith and E. Szathmáry, The major transitions in evolution. New York, NY, USA: Oxford University Press, 1997.
  2. C. Ofria and C. O. Wilke, “Avida: A software platform for research in computational evolutionary 
biology,” Journal of Artificial Life, vol. 10, pp. 191–229, 2004.
  3. H. J. Goldsby, N. Serra, F. Dyer, B. Kerr, and C. Ofria, “The evolution of temporal polyethism,” Artificial Life, vol. 13, pp. 178–185, 2012.
  4. H. J. Goldsby, D. B. Knoester, C. Ofria, and B. Kerr, “The evolutionary origin of somatic cells under the dirty work hypothesis,” PLoS Biol, vol. 12, no. 5:e1001858, 2014.
  5. H. J. Goldsby, A. Dornhaus, B. Kerr, and C. Ofria, “Task-switching costs promote the evolution of division of labor and shifts in individuality,” Proceedings of the National Academy of Sciences, vol. 109, no. 34, pp. 13686–13691, 2012.
Posted in BEACON Researchers at Work | Tagged , , , , , | Leave a comment

Evolution’s Pet Cemeteries: museum collections are haunted by ghosts of natural selection

Posted on February 3, 2016 by hageyt

This post is by MSU Postdoc Eben Gering.

Dr. Helen James. The Smithsonian bird collection Curator in Charge

Several of my most exciting workdays last year were spent rummaging through drawers of dead birds. I was on the hunt for a few dozen of the 640,000 specimens that makeup the world’s third largest bird collection. This collection resides within the Smithsonian National Museum of Natural History, but one could also say that it “lives” in the minds of the investigators who study it… Because there, like the cast of Milan Trenc’s A Night at the Museum, evolutionary processes spring to life from the inanimate artifacts that life leaves behind.

A curious biologist could easily dedicate her professional life to studying and curating such a treasure trove as the Smithsonian bird collection, and during my visit there I met two remarkable women who have done precisely that. I had already made an appointment to meet with one of them, Dr. Helen James to ask what could be learned about the biology of feral chickens (the focus of my current research) from several dried specimens in the Smithsonian collection. I’ll save the details for another day; the short answer is: a lot!

Dr. Carla Dove, the Department of Vertebrate Zoology Program Manager at the Smithsonian

Dr. James is among the world’s leading avian paleontologists, a reputation that rests upon decades of extraordinarily resourceful work. From meticulous studies of tiny bits of bone, she has pieced together much of the previously unknown evolutionary history of Pacific Oceania’s birds. Dr. James’ capabilities and reputation are likely what led someone from another museum to drop by her office during our meeting. They had brought with them several 100 million year old feathers that had been trapped (in breathtakingly pristine condition), in a recently discovered Burmese amber deposit. Would I mind if Helen took our conversation over to the feather forensics lab to have a look? No… I would not mind. For me (and probably also for you, if you are reading this blog), this was the stuff of childhood dreams.

Aircraft can occasionally ingest birds into their engines.

As Dr. James, her guest, and a postdoc excitedly examined the ancient feathers and sought “matches” from more recently collected material in the collections, I had an opportunity to meet the manager of the feather forensics lab, Dr. Carla Dove. Dr. Dove explained to me that, while the ancient feathers were a treat to examine, her lab is usually occupied by the study of younger, less well-preserved samples. Specifically, her team dedicates much of their effort to identifying birds from the fragmentary evidence that survives birdstrikes. These sometimes fatal accidents occur when aircraft ingest birds into their engines. Knowing the species (and therefor the biology) of the birds involved helps air traffic controllers, pilots, and engineers anticipate and avoid them.

The Smithsonian has Big Bird feathers in their collection!

Studies of birdstrikes offer a very compelling and straightforward example of why natural history collections, and the taxonomic experts who maintain them, are so important. Unfortunately many scientists and citizens are losing sight of the value of scientific collections. In my opinion a visit to a natural history collection is a must for any evolutionary biologist, no matter what animal, vegetable, or pathway s/he studies. We owe the very discovery of evolution to Darwin’s analysis of bird collections, and they offer unlimited opportunities for future breakthroughs.

Before leaving the museum, I asked Dr. Dove if she could share any interesting facts about chicken feathers. It turns out, she informed me, that barbules (tiny feather substructures) in the group of birds to which chickens belong (Galliformes), are unique in possessing multiple ringed nodes whose function is still unclear. Since a chicken feather was not at hand, Dr. Dove invited me to examine a relative’s feathers under the microscope. Yes…

Galliformes ringed barbules

SMITHSONIAN HAS BIG BIRD FEATHERS IN THEIR COLLECTION! And so at the close of a perfect day, I was able to view the Galliformes ringed barbules (image 5) because, it turns out, the immortal and monotypic Big Bird shares surprising morphological homology with Meliagris, the turkey.

 

Posted in BEACON Researchers at Work | Tagged , , , , | Leave a comment

Publicizing Your Research

Posted on February 1, 2016 by hageyt

This post is by MSU Media Communications Manager/Spartan Science Storyteller Layne Cameron.

First of all, allow me to introduce myself. I’m Layne Cameron, the Spartan Science Storyteller.

Layne Cameron, Spartan Science Storyteller, shares a few tips on working with the media and publicizing research.

My job is to help publicize the research that is being conducted at MSU. I write about your work for MSUToday, but I also pitch my stories to media outlets – from the New York Times and National Geographic to the local National Public Radio affiliate – to get reporters to cover the great science that’s happening at MSU.

I’ve worked with quite a few researchers from BEACON. The research that you do there is compelling, and your ability to translate the science for a general audience is first-rate. So, for the people I’ve helped, please keep the stories coming my way. And for those I’ve yet to meet, I looking forward to hearing from you.

MSU scientists often ask where I get my story ideas. My answer is, “From you.” I’d like to hear from you when:

  • You land a large grant to fund your research
  • You publish a paper in a peer-reviewed journal
  • Hear news coverage that’s related to your research
  • When you snap a cool image that captures an aspect of your work
  • When you’re being interviewed for a story

In all of those situations, I can help publicize your work and coordinate media coverage and social media strategy.

When you land a grant or publish a paper, you’re often told: “YOU CAN’T TALK TO ANYONE ABOUT THIS!!” That’s true; you can’t talk the media until you have the OK from granting organization or the embargo lifts for your paper.

However, you can talk to me beforehand—I am a media relations manager NOT a reporter. I can meet with you to prepare talking points, news release, take photos, and rough out a media and social media strategy in advance of any grant or embargo. I will hold all of the work until the embargoes are lifted. (And with papers in journals such as Science, Nature and PNAS, I can even pitch the stories to a handful of trusted, national science reporters, who will then hold their stories until the appointed time.)

You don’t need to write a news release, either – that’s my job! Simply drop me an email or give me a call. Send me a copy of your grant application or paper with a layman’s description, and we’ll get started.

To give you a better idea of what I do, I’ve included a couple of links to stories that I’ve covered recently.

I’m always open to meet with anyone to discuss the ins and outs of working with the media. If you have an idea, let’s talk. Over coffee. In your lab. Wherever.

If you’re still unsure, feel free to cyber-stalk me on Twitter. I’m at SpartnSciStorytellr or @LayneCameron1. (I’ll take all the followers I can get!)

Posted in BEACON in the News, BEACONites | Tagged , , , | Leave a comment

Point Break: My experiment with dropping PowerPoint in a large lecture course

Posted on January 25, 2016 by hageyt

This post is by MSU Associate Professor Chris Waters.

A Change is necessary

I am the course administrator and sole instructor for the junior/senior level course “MMG 431:Microbial Genetics” of about 150 students at Michigan State University. Entering my seventh year of teaching this course in the Fall of 2015, I began to reevaluate my approach. Previously, I had implemented a fairly standard lecture course with PowerPoint slides interspaced with active learning “think:pair:share” exercises that utilized iClickers for students participation. The students were given the PowerPoint slides before class, and these were intended to be an outline for further note taking during class.

But each year the complaints of the students were the same: “too much material…”, “the instructor goes too fast…”

From my perspective, I was frustrated that half the class was clearly not paying attention, but rather felt that the 50 minutes of class time were ideal for catching up on Facebook, Instagram, or their favorite electronic distraction.

To address both of these problems, I decided to stop using PowerPoint (with a few exceptions for complicated structures or diagrams) and rather present information in a “chalk-talk” style where the students take notes as I write everything from scratch. Do any of you remember those days, say from the beginning of time until the year 2,000, when you came to class with a blank notebook, two pencils, and a readiness to take notes! Here is how this experiment unfolded last Fall…

Choosing a format

Although I refer to this style as chalk-talk, no chalk was actually harmed during the making of my course. Rather, I decided to use my touch-screen laptop and project what I was writing on two screens in my lecture hall. This has several advantages including more color options, the ability to switch between formats (the rare PowerPoint slide, online videos, etc.), and it allowed me to save every lecture so I could reference exactly what I had presented. I explored different formats but eventually decided to use OneNote 2013. OneNote allows you to have an infinitely large screen that you can zoom in/out, it has many colors and pen styles, and it nicely organized for keeping track of each lecture as a separate entry. To prepare for class, I would basically transcribe my old slides into handwritten notes in a bound notebook, and use this to give my lecture. Being my seventh year, I know the material inside and out so this was easy for me to do (it would be much harder starting out with a new course).

Fig. 1. The classic Hershey-Chase experiment. Sidenote: Alfred Hershey obtained his PhD from the Microbiology Dept. at MSU!

Here is an example of my description of the classic Hershey-Chase experiment utilizing phage to elucidate DNA as the hereditary material (Fig. 1). It doesn’t look like much, but remember I am drawing this from scratch explaining what is happening as I go along so hopefully the student’s versions would be filled with all kinds of additional comments that I only verbally present.

How did it work?

At the start of the semester I was quite nervous about pulling this off, especially given my tendency for sloppy hand-writing. But I quickly became comfortable with the new approach, and I grew to love teaching this way. I was more able to emphasize important points and comments, and as one student mentioned, “it’s very interesting to see the way your thought processes unfold when you provide illustrations”. Another fantastic aspect of this method is that is allowed me much more freedom to take the class in new directions or present a new idea on the spot. Unlike with PowerPoint, I was not fixed to a given order, and the students did not know what was coming next. Many times during the course I improvised in ways which I had not been able to do before. I loved the freedom! It also allowed me to better query the students and report their responses. For example, I could ask “What are some of the potential uses and drawbacks of CRISPRs?” and the students were not able to merely reference the next slide in that day’s lecture.

Importantly, the students were all “locked-in” during class! The unfocused, inattentive student had magically vanished. Check for solving my past complaint. And, let me tell you, it takes a heck of a lot longer to draw something out rather than flash up a PP slide. Not surprisingly, I quickly got behind my normal course schedule and had to adjust the information that I presented to cut details or additional examples. I would estimate I that I trimmed 35% of the material that I had given in the prior year, and students complained much less on the amount of content and lecture speed. Check for addressing the student’s complaints.

Quantitative data: student response and effectiveness

It is all well and good that I liked teaching in a chalk-talk style, but more importantly how did the students respond to it and was it effective?

Fig. 2. Student’s numerical review scores 2013-2015.

To address these question, I analyzed the student’s evaluations and course scores from 2013-2015. This entails the three years that I have been the sole instructor of MMG431. Scores are ranked from 1 to 5 with 1 being the best and 5 being the worst. The questions can be grouped into 6 categories such as “Instructor Involvement”, “Student Interest”, etc. Although 2013 and 2014 were fairly equivalent (both “normal” PowerPoint lectures), every single category improved last Fall with the implementation of the chalk-talk format (Fig. 2)! The biggest gains occurred in “Course Demands” and “Student Enjoyment”, historically the two weakest categories.

Fig. 3-Categorizing specific student comments.

In addition to these numerical scores, I also evaluated specific student comments from the last three years, grouping them into whether they had an overall negative, positive, or neutral view of the course. The results were striking with a huge increase in positive responses from 18-30% to almost 70% associated with a corresponding decrease in negative responses from 65-66% to 21% (Fig. 3).

Clearly the new approach resonated with the students. Examples of specific comments that I received were “It is extremely easy to take notes”, “I think the changes made on how to teach this course were super effective”, and “I wish all my classes would go back to teaching instead of reading material off of slides”. And the most surprising comment, which I have never seen a similar one in 7 years, actually wanted class to be longer: “If only classes ran 1 hour and 20 minutes 3 times a week instead of 50 minutes so we could cover more.”

Fig. 4. Student’s unadjusted final scores.

But how effective was the chalk-talk style? The hope is that they retain more information. To answer this question, I compared the final, unadjusted grades from 2013-2015. These are the raw grades at the end of the semester without any adjustment for overall course difficulty (Fig. 4). Although there was a small i

ncrease in the 4.0 group, it was not as dramatic as other groups, suggesting to me that the top students will do well regardless of what format you use. But the largest increases were seen in the 3.5 and 3.0 categories. These are students who likely would have in the past received 2.0 or below who were able to be more successful with the new style. Most dramatically was the decrease in the number of failing students, which has been one of my long-term goals for this course.

Getting to the point

The data indicate that for myself and my course, the chalk-talk style was highly superior to traditional PowerPoint lectures. There is no question that I will continue teaching in this manner. However, there is clearly still room for improvement.

A number of complaints focused on legibility issues. This comes from my own natural propensity for chicken-scratching and my laptop format, which was somewhat unnatural to write/draw. Before next Fall, I will switch to a tablet format which will hopefully improve these issues. More supplemental slides with complex diagrams was also a common request, and I believe I can make more of these available. Some students had a hard time keeping up with note taking, and I anticipate using two devices next year alternating between two screens so that material is displayed longer. In my opinion, many of them just do not have the experience or skill to take good notes-they have never had to do it before. Perhaps a primer on good note taking would be warranted. Other students such as non-native speakers or students with mental health issues expressed frustration that they would often miss points in class and it was difficult to make them up since I did not post lecture material. I plan to record my lectures and make an audio version available, and I am debating about whether to post lecture notes after the class is over. Any thoughts in the comments section about posting lecture notes are appreciated!

I consider “Point Break” a highly successful experiment that met all of my desired outcomes. On a fundamental level, presenting in a chalk-talk style just made teaching more fun. Consider me a convert.

Posted in BEACONites, Education | Tagged , , | Leave a comment