Ryan Gutenkunst Theoretical Biology and Biophysics & Center for Nonlinear Studies Los Alamos National Laboratory ryang@lanl.gov ryan@gutenkunst.org

Peer reviewed publications

1. Genome variation of rice under domestication
   X Xu°, S Ge°, JD Jensen°, RN Gutenkunst°, J Li°, X Li°, L Fang, L Huang, K Ye, W He, G Zhang, X Zheng, F Zhang, F Hu, Y Li, C Yu, Y Guo, D Li, K Kristiansen, X Zhang, J Wang, R Nielsen, J Wang, W Wang (°contributed equally)
   submitted; LA-UR 09-07280
2. Proteins with greater influence on network dynamics evolve more slowly but are not more essential
   RN Gutenkunst
   submitted; arXiv:0909.2889; LA-UR 09-05782; toggle abstract

   A fundamental question for evolutionary biology is why rates of evolution vary dramatically between proteins. Perhaps surprisingly, it is controversial how much a protein's functional importance affects its rate of evolution. In most studies, functional importance has been measured on the coarse scale of protein knock-outs, while evolutionary rate has been measured on the fine scale of amino acid substitutions. Here we introduce dynamical influence, a finer measure of protein functional importance. To measure dynamical influence, we first use detailed biochemical models of particular reaction networks to measure the influence of each reaction rate constant on network dynamics. We then define the dynamical influence of a protein to be the average influence of the rate constants for all reactions it is involved in.

   Using models of a dozen biochemical systems and sequence data from vertebrates, we show that dynamical influence and evolutionary rate are negatively correlated; proteins with greater dynamical influence evolve more slowly. We also show that proteins with greater dynamical influence are not more likely to be essential. This suggests that there are many cellular reactions whose presence is essential for life, but whose quantitative rate is relatively unimportant to fitness. We also provide evidence that the effect of dynamical influence on evolutionary rate is independent of protein expression level, expression specificity, gene compactness, and reaction degree. Dynamical influence offers a finer view of functional importance, and our results suggest that focusing on essentiality may have previously led to an underestimation of the role function plays in protein evolution.

3. Adaptive mutation in a geometrical model of chemotype evolution
   RN Gutenkunst, JP Sethna
   submitted; arXiv:0712.3240; toggle abstract

   The distribution of fitness effects of adaptive mutations remains poorly understood, both empirically and theoretically. Most recent theoretical work on the subject has focused on either the genotype, the fundamental information encoding of the organism, or on the phenotype, the emergent properties of the organism. Here we study a version of Fisher's geometrical model formulated in terms of the chemotype, the physical properties of the biomolecules (such as binding affinities and reaction rates) that connect the genotype and the phenotype. We assume that any given mutation affects only one element of the chemotype (i.e. minimal pleiotropy). Our model generically predicts striking cusps in the distribution of fitness effects of fixed mutations and predicts that a single element of the chemotype should comprise all mutations at the high fitness end of that distribution. Using extreme value theory, we show that the most fit pair of these cusps are typically well-separated, even when the chemotype consists of thousands of elements, implying that the effects we predict should be observable in practical experiments. More broadly, our work demonstrates that new insight can be gain by viewing evolution through the chemotype, the level intermediate between the genotype and the phenotype.

4. Human population genetics and genomics
   S Ramachandran, H Tang, RN Gutenkunst, CD Bustamante
   in M Speicher et al., editors, Vogel and Motulsky's Human Genetics: Problems and Approaches, Springer Verlag (in press); LA-UR 09-05785; toggle abstract

   Recent developments in sequencing technology have created a flood of new data on human genetic variation, and this data has yielded new insights into human population structure. Here we review what both early and more recent studies have taught us about human population structure and history. Early studies showed that most human genetic variation occurs within populations rather than between them, and that genetically related populations often cluster geographically. Recent studies based on much larger data sets have recapitulated these observations, but have also demonstrated that high-density genotyping allows individuals to be reliably assigned to their population of origin. In fact, for admixed individuals, even the ancestry of particular genomic regions can often be reliably inferred. Recent studies have also offered detailed information about the composition of specific populations from around the world, revealing how history has shaped their genetic makeup. We also briefly review quantitative models of human genetic history, including the role natural selection has played in shaping human genetic variation.

5. Inferring the joint demographic history of multiple populations from multidimensional SNP data
   RN Gutenkunst, RD Hernandez, SH Williamson, CD Bustamante
   PLoS Genetics 5(10):e1000695 (2009); arXiv:0909.0925; publisher copy; local copy; local supporting info; LA-UR: 09-03343; toggle abstract

   Demographic models built from genetic data play important roles in illuminating prehistorical events and serving as null models in genome scans for selection. We introduce an inference method based on the joint frequency spectrum of genetic variants within and between populations. For candidate models we numerically compute the expected spectrum using a diffusion approximation to the one-locus, two-allele Wright-Fisher process, involving up to three simultaneous populations. Our approach is a composite likelihood scheme, since linkage between neutral loci alters the variance but not the expectation of the frequency spectrum. We thus use bootstraps incorporating linkage to estimate uncertainties for parameters and significance values for hypothesis tests. Our method can also incorporate selection on single sites, predicting the joint distribution of selected alleles among populations experiencing a bevy of evolutionary forces, including expansions, contractions, migrations, and admixture. We model human expansion out of Africa and the settlement of the New World, using 5 Mb of noncoding DNA resequenced in 68 individuals from 4 populations (YRI, CHB, CEU, and MXL) by the Environmental Genome Project. We infer divergence between West African and Eurasian populations 140 thousand years ago (95% confidence interval: 40-270 kya). This is earlier than other genetic studies, in part because we incorporate migration. We estimate the European (CEU) and East Asian (CHB) divergence time to be 23 kya (95% c.i.: 17-43 kya), long after archeological evidence places modern humans in Europe. Finally, we estimate divergence between East Asians (CHB) and Mexican-Americans (MXL) of 22 kya (95% c.i.: 16.3-26.9 kya), and our analysis yields no evidence for subsequent migration. Furthermore, combining our demographic model with a previously estimated distribution of selective effects among newly arising amino acid mutations accurately predicts the frequency spectrum of nonsynonymous variants across three continental populations (YRI, CHB, CEU).

6. Targets of balancing selection in the human genome
   AM Andrés, MJ Hubisz, A Indap, D Torgerson, J Degenhardt, AR Boyko, RN Gutenkunst, TJ White, ED Green, CD Bustamante, AG Clark, R Nielsen
   Molecular Biology and Evolution 26(12):2755-2764 (2009); publisher copy; local copy; toggle abstract

   Balancing selection is potentially an important biological force for maintaining advantageous genetic diversity in populations, including variation that is responsible for long-term adaptation to the environment. By serving as a means to maintain genetic variation, it may be particularly relevant to maintaining phenotypic variation in natural populations. Nevertheless, its prevalence and specific targets in the human genome remain largely unknown. We have analyzed the patterns of diversity and divergence of 13,400 genes in two human populations using an unbiased single-nucleotide polymorphism data set, a genome-wide approach, and a method that incorporates demography in neutrality tests. We identified an unbiased catalog of genes with signatures of long-term balancing selection, which includes immunity genes as well as genes encoding keratins and membrane channels; the catalog also shows enrichment in functional categories involved in cellular structure. Patterns are mostly concordant in the two populations, with a small fraction of genes showing population-specific signatures of selection. Power considerations indicate that our findings represent a subset of all targets in the genome, suggesting that although balancing selection may not have an obvious impact on a large proportion of human genes, it is a key force affecting the evolution of a number of genes in humans.

7. Darwinian and demographic forces affecting human protein coding genes
   R Nielsen, MJ Hubisz, I Hellmann, D Torgerson, AM Andrés, A Albrechtsen, R Gutenkunst, MD Adams, M Cargill, X Hu, A Boyko, A Indap, CD Bustamante, AG Clark
   Genome Research 19:838-849 (2009); publisher copy; local copy; toggle abstract

   Past demographic changes can produce distortions in patterns of genetic variation that can mimic the appearance of natural selection unless the demographic effects are explicitly removed. Here we fit a detailed model of human demography that incorporates divergence, migration, admixture, and changes in population size to directly sequenced data from 13,400 protein coding genes from 20 European-American and 19 African-American individuals. Based on this demographic model, we use several new and established statistical methods for identifying genes with extreme patterns of polymorphism likely to be caused by Darwinian selection, providing the first genome-wide analysis of allele frequency distributions in humans based on directly sequenced data. The tests are based on observations of excesses of high frequency–derived alleles, excesses of low frequency–derived alleles, and excesses of differences in allele frequencies between populations. We detect numerous new genes with strong evidence of selection, including a number of genes related to psychiatric and other diseases. We also show that microRNA controlled genes evolve under extremely high constraints and are more likely to undergo negative selection than other genes. Furthermore, we show that genes involved in muscle development have been subject to positive selection during recent human history. In accordance with previous studies, we find evidence for negative selection against mutations in genes associated with Mendelian disease and positive selection acting on genes associated with several complex diseases.

8. Global distribution of genomic diversity underscores rich complex history of continental human populations
   A Auton, K Bryc, AR Boyko, KE Lohmueller, J Novembre, A Reynolds, A Indap, MH Wright, J Degenhardt, RN Gutenkunst, KS King, MR Nelson, CD Bustamante
   Genome Research 19:795-803 (2009); publisher copy; local copy; toggle abstract

   Characterizing patterns of genetic variation within and among human populations is important for understanding human evolutionary history and for careful design of medical genetic studies. Here, we analyze patterns of variation across 443,434 single nucleotide polymorphisms (SNPs) genotyped in 3845 individuals from four continental regions. This unique resource allows us to illuminate patterns of diversity in previously under-studied populations at the genome-wide scale including Latin America, South Asia, and Southern Europe. Key insights afforded by our analysis include quantifying the degree of admixture in a large collection of individuals from Guadalajara, Mexico; identifying language and geography as key determinants of population structure within India; and elucidating a north-south gradient in haplotype diversity within Europe. We also present a novel method for identifying long-range tracts of homozygosity indicative of recent common ancestry. Application of our approach suggests great variation within and among populations in the extent of homozygosity, suggesting both demographic history (such as population bottlenecks) and recent ancestry events (such as consanguinity) play an important role in patterning variation in large modern human populations.

9. Sloppiness, robustness, and evolvability in systems biology
   BC Daniels, YJ Chen, JP Sethna, RN Gutenkunst, CR Myers
   Current Opinion in Biotechnology 19:389-395 (2008); arXiv:0805.2628; publisher copy; local copy; local supporting info; toggle abstract

   The functioning of many biochemical networks is often robust - remarkably stable under changes in external conditions and internal reaction parameters. Much recent work on robustness and evolvability has focused on the structure of neutral spaces, in which system behavior remains invariant to mutations. Recently we have shown that the collective behavior of multiparameter models is most often sloppy: insensitive to changes except along a few "stiff" combinations of parameters, with an enormous sloppy neutral subspace. Robustness is often assumed to be an emergent evolved property, but the sloppiness natural to biochemical networks offers an alternative nonadaptive explanation. Conversely, ideas developed to study evolvability in robust systems can be usefully extended to characterize sloppy systems.

10. Variational method for estimating the rate of convergence of Markov Chain Monte Carlo algorithms
    FP Casey, JJ Waterfall, RN Gutenkunst, CR Myers, JP Sethna
    Physical Review E 78:046704 (2008); arXiv:physics/0609001; publisher copy; local copy; toggle abstract

    We demonstrate the use of a variational method to determine a quantitative lower bound on the rate of convergence of Markov chain Monte Carlo (MCMC) algorithms as a function of the target density and proposal density. The bound relies on approximating the second largest eigenvalue in the spectrum of the MCMC operator using a variational principle and the approach is applicable to problems with continuous state spaces. We apply the method to one dimensional examples with Gaussian and quartic target densities, and we contrast the performance of the random walk Metropolis-Hastings algorithm with a "smart" variant that incorporates gradient information into the trial moves, a generalization of the Metropolis adjusted Langevin algorithm. We find that the variational method agrees quite closely with numerical simulations. We also see that the smart MCMC algorithm often fails to converge geometrically in the tails of the target density except in the simplest case we examine, and even then care must be taken to choose the appropriate scaling of the deterministic and random parts of the proposed moves. Again, this calls into question the utility of smart MCMC in more complex problems. Finally, we apply the same method to approximate the rate of convergence in multidimensional Gaussian problems with and without importance sampling. There we demonstrate the necessity of importance sampling for target densities which depend on variables with a wide range of scales.

11. Universally sloppy parameter sensitivities in systems biology
    RN Gutenkunst, JJ Waterfall, FP Casey, KS Brown, CR Myers, JP Sethna
    PLoS Computational Biology 3(10):e189 (2007); arXiv:q-bio.QM/0701039; publisher copy; local copy; local supporting info; toggle abstract
    • Featured in a one-page article: A Dance, "Sloppy" Systems Biology, Biomedical Computation Review Winter:2 (2007/08); publisher copy; local copy
    • "Exceptional" evaluation on Faculty of 1000: Biology.

    Quantitative computational models play an increasingly important role in modern biology. Such models typically involve many free parameters, and assigning their values is often a substantial obstacle to model development. Directly measuring in vivo biochemical parameters is difficult, and collectively fitting them to other experimental data often yields large parameter uncertainties. Nevertheless, in earlier work we showed in a growth-factor-signaling model that collective fitting could yield well-constrained predictions, even when it left individual parameters very poorly constrained. We also showed that the model had a "sloppy" spectrum of parameter sensitivities, with eigenvalues roughly evenly distributed over many decades. Here we use a collection of models from the literature to test whether such sloppy spectra are common in systems biology. Strikingly, we find that every model we examine has a sloppy spectrum of sensitivities. We also test several consequences of this sloppiness for building predictive models. In particular, sloppiness suggests that collective fits to even large amounts of ideal time-series data will often leave many parameters poorly constrained. Tests over our model collection are consistent with this suggestion. This difficulty with collective fits may seem to argue for direct parameter measurements, but sloppiness also implies that such measurements must be formidably precise and complete to usefully constrain many model predictions. We confirm this implication in our growth-factor-signaling model. Our results suggest that sloppy sensitivity spectra are universal in systems biology models. The prevalence of sloppiness highlights the power of collective fits and suggests that modelers should focus on predictions rather than on parameters.

12. Extracting falsifiable predictions from sloppy models
    RN Gutenkunst, FP Casey, JJ Waterfall, CR Myers, JP Sethna
    in G. Stolovitsky, A. Califano, J. Collins, editors, Reverse Engineering Biological Networks: Opportunities and Challenges in Computational Methods for Pathway Inference
    Annals of the New York Academy of Sciences 1115:203-211 (2007); arXiv:0704.3049; publisher copy; local copy; toggle abstract

    Successful predictions are among the most compelling validations of any model. Extracting falsifiable predictions from nonlinear multiparameter models is complicated by the fact that such models are commonly sloppy, possessing sensitivities to different parameter combinations that range over many decades. Here we discuss how sloppiness affects the sorts of data that best constrain model predictions, makes linear uncertainty approximations dangerous, and introduces computational difficulties in Monte-Carlo uncertainty analysis. We also present a useful test problem and suggest refinements to the standards by which models are communicated.

13. Optimal experimental design in an epidermal growth factor receptor signalling and down-regulation model
    FP Casey, D Baird, Q Feng, RN Gutenkunst, JJ Waterfall, CR Myers, KS Brown, RA Cerione, JP Sethna
    IET Systems Biology 1(3):190-202 (2007); arXiv:q-bio.MN/0610024; publisher copy; local copy; toggle abstract

    We apply the methods of optimal experimental design to a differential equation model for epidermal growth factor receptor signalling, trafficking and down-regulation. The model incorporates the role of a recently discovered protein complex made up of the E3 ubiquitin ligase, Cbl, the guanine exchange factor (GEF), Cool-1 (Β-Pix) and the Rho family G protein Cdc42. The complex has been suggested to be important in disrupting receptor down-regulation. We demonstrate that the model interactions can accurately reproduce the experimental observations, that they can be used to make predictions with accompanying uncertainties, and that we can apply ideas of optimal experimental design to suggest new experiments that reduce the uncertainty on unmeasurable components of the system.

14. Python unleashed on systems biology
    CR Myers, RN Gutenkunst, JP Sethna
    Computing in Science and Engineering 9(3):34-37 (2007); arXiv:0704.3259; publisher copy; local copy; toggle abstract

    Researchers at Cornell University have built an open source software system to model biomolecular reaction networks. SloppyCell is written in Python and uses third-party libraries extensively, but it also does some fun things with on-the-fly code generation and parallel programming.

15. Inferring resource distributions from Atlantic bluefin tuna movements: an analysis based on net displacement and length of track
    R Gutenkunst, N Newlands, M Lutcavage, L Edelstein-Keshet
    Journal of Theoretical Biology 245(2):243-257 (2007); publisher copy; local copy; toggle abstract

    We use observed movement tracks of Atlantic bluefin tuna in the Gulf of Maine and mathematical modeling of this movement to identify possible resource patches. We infer bounds on the overall sizes and distribution of such patches, even though they are difficult to quantify by direct observation in situ. To do so, we segment individual fish tracks into intervals of distinct motion types based on the ratio of net displacement to length of track (ΔD/ΔL) over a time window Δt. To find the best segmentation, we optimize the fit of a random-walk movement model to each motion type. We compare results from two distinct movement models: biased turning and biased speed, to check the model-dependence of our inferences, and find that uncertainty in choice of movement model dominates the uncertainties of our conclusions. We find that our data are best described using two motion types: "localized" (ΔD/ΔL small) and "long-ranged" (ΔD/ΔL large). The biased turning model leads to significantly better resolution of localized movement intervals than the biased speed model. We hypothesize that localized movement corresponds to exploitation of resource patches. Comparison with visual behavior observations made during tracking suggests that many inferred intervals of localized motion do indeed correspond to feeding activity. From our analysis, we estimate that, on average, bluefin tuna in the Gulf of Maine encounter a resource patch every 2 h, that those patches have an average radius of 0.7-1.2 km, and that, overall, there are at most 5-9 such patches per 100 km² in the region studied.

16. Sloppy-model universality class and the Vandermonde matrix
    JJ Waterfall, FP Casey, RN Gutenkunst, KS Brown, CR Myers, PW Brouwer, V Elser, JP Sethna
    Physical Review Letters 97:150601 (2006); arXiv:cond-mat/0605387; publisher copy; local copy; toggle abstract
    

    In a variety of contexts, physicists study complex, nonlinear models with many unknown or tunable parameters to explain experimental data. We explain why such systems so often are sloppy: the system behavior depends only on a few "stiff" combinations of the parameters and is unchanged as other "sloppy" parameter combinations vary by orders of magnitude. We observe that the eigenvalue spectra for the sensitivity of sloppy models have a striking, characteristic form with a density of logarithms of eigenvalues which is roughly constant over a large range. We suggest that the common features of sloppy models indicate that they may belong to a common universality class. In particular, we motivate focusing on a Vandermonde ensemble of multiparameter nonlinear models and show in one limit that they exhibit the universal features of sloppy models.

17. An introduction to signal extraction in interferometric gravitational wave detectors
    ED Black, RN Gutenkunst
    American Journal of Physics 71(4):365-378 (2003); publisher copy; local copy; toggle abstract

    In the very near future gravitational wave astronomy is expected to become a reality, giving us a completely new tool for exploring the universe around us. We provide an introduction to how interferometric gravitational wave detectors work, suitable for students entering the field and teachers who wish to cover the subject matter in an advanced undergraduate or beginning graduate level course

18. Extracting light from water: Sonoluminescence
    R Gutenkunst
    Caltech Undergraduate Research Journal 2(1):16-22 (2002); local copy; toggle abstract

    Water is one of the most abundant compounds on Earth. No known form of life can survive without it, and for a human, a day without water is a day of suffering. Water is such a familiar element in everyday life that one might assume that it is completely understood. Unfortunately, that is not true. Our ignorance is well illustrated by sonoluminescence, the astounding ability of water to concentrate energy from sound waves into flashes of light. After over a decade of intense study, the scientific community is still unable to provide a complete explanation for this phenomenon, but its work has pushed the frontiers of fluid mechanics and expanded our understanding of physics at high pressures and temperatures.

Software

• ∂a∂i - Rapid and powerful diffusion method for fitting demographic models to genetic polymorphism data.
• SloppyCell - SBML-compatible environment for building, fitting, and analyzing biochemical models, with an emphasis on parameter sensitivity.

Other science publications

1. Sloppiness, Modeling, and Evolution in Biochemical Networks
   RN Gutenkunst
   Ph.D. thesis, Cornell University (2007); university copy; hyperlinked local copy; toggle abstract

   The wonderful complexity of livings cells cannot be understood solely by studying one gene or protein at a time. Instead, we must consider their interactions and study the complex biochemical networks they function in.

   Quantitative computational models are important tools for understanding the dynamics of such biochemical networks, and we begin in Chapter 2 by showing that the sensitivities of such models to parameter changes are generically "sloppy", with eigenvalues roughly evenly spaced over many decades. This sloppiness has practical consequences for the modeling process. In particular, we argue that if one's goal is to make experimentally testable predictions, sloppiness suggests that collectively fitting model parameters to system-level data will often be much more efficient that directly measuring them.

   In Chapter 3 we apply some of the lessons of sloppiness to a specific modeling project involving in vitro experiments on the activation of the heterotrimeric G protein transducin. We explore how well time-series activation experiments can constrain model parameters, and we show quantitatively that the T177A mutant of transducin exhibits a much slower rate of rhodopsin-mediated activation than the wild-type.

   All the preceding biochemical modeling work is performed using the SloppyCell modeling environment, and Chapter 4 briefly introduces SloppyCell and some of the analyses it implements. Additionally, the two appendices of this thesis contain preliminary user and developer documentation for SloppyCell.

   Modelers tweak network parameters with their computers, and nature tweaks such parameters through evolution. We study evolution in Chapter 5 using a version of Fisher's geometrical model with minimal pleiotropy, appropriate for the evolution of biochemical parameters. The model predicts a striking pattern of cusps in the distribution of fitness effects of fixed mutations, and using extreme value theory we show that the consequences of these cusps should be observable in feasible experiments.

   Finally, this thesis closes in Chapter 6 by briefly considering several topics: sloppiness in two non-biochemical models, two technical issues with building models, and the effect of sloppiness on evolution beyond the first fixed mutation.

Selected presentations

• Proteins with large influence on network dynamics evolve slowly (320 KB pdf)
  Poster: Cold Spring Harbor Symposium, Evolution: the molecular landscape (05/28/09, LA-UR: 09-03251)
  
• Inferring Human History from the Joint Site-Frequency Spectrum (1.1 MB pdf)
  Poster: Cornell Postdoc Research Day, winner of a Best Poster Presentation award (05/08/08)
  
• Universal Sloppiness in Complex Biology: Implications for Modeling and Evolution (3.2 MB pdf)
  Talk: Computational Biology Student Seminar (02/09/07)
• Probing the effects of specific transducin mutations via computational kinetic modeling (492 KB pdf)
  Poster: Keystone Symposium on G Protein Coupled Receptors (02/13/06)

Photography publications

1. Los Gatos Holds 52nd Parade
   A Dallas, R Gutenkunst (photographer)
   Los Gatos Observer December 9 (2008); link
2. Black Tusk: Črna gora v Britanski Kolumbiji
   M Željko, R Gutenkunst (photographer)
   Planinski Vestnik May:22 (2004); local copy (sadly, poor photo quality in pdf)

• For more photos, see Shannon and my flickr page.

Contact

• Email: ryang@lanl.gov
• Office: TA-3, Bldg 1690, Room 117
• Phone: (505) 665-2367
• Fax: (505) 665-7652
• Mailing address:
  MS B258
  Los Alamos National Laboratory
  Los Alamos, NM 87545