Seminars are held Tuesdays and Thursdays in POB 6.304 from 3:30-5:00 pm, unless otherwise noted. Speakers include scientists, researchers, visiting scholars, potential faculty, and ICES/UT Faculty or staff. Everyone is welcome to attend. Refreshments are served at 3:15 pm.
Metabolism is defined as the full complement of chemical transformations in living systems. In many biotechnological applications an organism must be endowed with additional metabolic capabilities. To support this goal, we introduced computational tools for constructing thermodynamically feasible, carbon and energy efficient, overall conversion stoichiometries by globally assessing all possible co-reactant/products combinations. Protein engineering, either de novo or through directed evolution, can expand upon the parts-list available for constructing pathways by re-purposing existing enzymes for novel conversions. However, minimizing the use of novel steps and combining them with existing reactions has remained so far elusive. In response to this unmet need, we developed novoStoic, which allows for seamlessly blending known reactions with de novo steps to construct atom and energy balanced pathways using reaction atom mapping information. Atom tracking through reactions also enables metabolic flux elucidation at a genome-scale revealing how the assumptions implied by core metabolic models may propagate in the inference of internal metabolic fluxes. Finally, we will highlight ongoing efforts that make use of multiple flux datasets of deletion mutants to robustly parameterize kinetic models that approach genome-scale for E. coli and other microbes. We will conclude by offering insight and lessons gained from the application of these tools on a variety of bioproduction challenges.
Costas D. Maranas is the Donald B. Broughton Professor in the Department of Chemical Engineering at The Pennsylvania State University. He received a Diploma in Chemical Engineering at the Aristotle University, Greece in 1990 and a Ph.D. in Chemical Engineering from Princeton University in 1995. He has been in the faculty of the department of Chemical Engineering at Penn State since 1995. He is the recipient of the Allan P. Colburn Award for Excellence in Publications by a Young Member of AIChE (2002), the Outstanding Young Investigator Award of the Computing and Systems Technology AIChE Division (2006), the S.V. Sotirchos Lectureship at 6th Panhellenic Chemical Engineering Conference (2007), the Penn State Engineering Alumni Society (PSEAS) Premier Research Award (2016) and Outstanding Research Award in (2012). He is a member of a number of journal Editorial Boards including PLOS Computational Biology, BMC Systems Biology, Biotechnology Journal and Metabolic Engineering. He is a Fellow of the American Institute of Medical and Biological Engineering (AIMBE). He is a member of advisory/steering committees for PNNL/EMSL and EcoCyc and the “Use Inspired Research” Lead in the Center for Bioenergy Innovation (CBI) DOE center.
The C. Maranas group develops and deploys computational framework informed by systems engineering and mathematical optimization to understand, analyze and redesign metabolism and proteins. Research interests include: Computational protein design; enzyme and antibody engineering; design of protein pores for bioseparations; reconstruction, curation and analysis of metabolic networks; computational strain design and synthetic biology; metabolism of photosynthetic organisms; metabolism of obligatory anaerobes; modeling of microbial communities; optimization theory and algorithms. He has co-authored over 180 refereed journal publications including a textbook on “Optimization Methods in Metabolic Networks” (2016). He has supervised 32 PhD theses with many group alumni occupying leading positions in industry and academia. http://maranasgroup.com
Cryo-electron microscopy (cryo-EM) is an imaging technique to determine the shape of 3-D macromolecules from massive datasets of noisy 2-D projections. Recognized by the 2017 Chemistry Nobel Prize, cryo-EM captures molecules in their native states, in contrast to traditional imaging modalities. The tradeoff is that the resulting inverse problem has rich mathematical and computational challenges, including extreme noise, unknown orientations, big data and conformational heterogeneity.
This lecture presents computational mathematics that has arisen from cryo-EM. In particular, I will discuss a general framework for
estimation under compact group actions, that connects information theory and group invariant theory. Other perspectives from
computational algebra also appear, including robust solvers for certain large-scale polynomial systems and fast methods for tensor
rank decomposition. These ideas are brought together in a moment-based approach for ab initio cryo-EM reconstruction, in which the computational complexity is controlled mostly by the desired resolution rather than the size of the dataset.
The development of replacement heart valve usually requires expensive and time-consuming large animal models. While murine models have many advantages in both speed and quantities, also allowing for the use of genetic knockout techniques, the very small size of the heart valve (~1mm) greatly limits their use and research. Particularly, it is still not clear about the unloaded shape of murine heart valves and their mechanical properties.
In this talk, I will present an integrated imaging/computational method for studying murine heart valves. In the first part, I'll show our up-to-date analysis on high-resolution micro-CT images of a collection of nine murine heart valves. While different valves are fixed under different pressure level from 0 to 30mmHg, remarkably, we found a scaling law that the length of free edge changes with pressure in proportion to the circumference of the entire valve. In the second part, I'll discuss a computational pipeline to analyze the imaging data. The approach is based on isogeometric analysis and treats the valve as a 2D shell with leaflet-leaflet contact formulated by a volumetric potential. Using the scaling law as a geometric constraint, we systematically studied how unloaded valve geometry and material properties affect valve shape under uniform pressure. Therefore, by optimizing against the micro-CT images, we estimate, for the first time, unloaded murine pulmonary heart valve shape and mechanical properties. Collectively, I aim to demonstrate the capability of our integrated imaging/computational method and the potential of using murine animal models in pulmonary heart valve research and regenerative approaches for replacement heart valves.
Dr. Feng is a postdoc fellow at Willerson Center for Cardiovascular Modeling and Simulation at the University of Texas at Austin. He earned his PhD in Mechanical Engineering from Cornell University (2016) in the area of solid mechanics. Dr. Feng's research interests are focused on computational biomechanics in cardiovascular system, computational oncology, computational cell mechanics and cell traction microscopy.
Speaker Affiliation: Associate Professor, Applied Mathematics, Stony Brook University, and Affiliate Member, Laufer Center for Physical and Quantitative Biology, Institute of Chemical Biology & Drug Discovery, and Institute of Advance Computational Sciences.
The protoplasmic revolution provided the blueprint for the networks of molecular interactions in the cell, however full understanding of how molecules interact comes only from three-dimensional structures. Despite recent progress in structure determination of individual proteins using X-ray or NMR, structures of complexes remains difficult to obtain. Additionally, modulating protein interactions for therapeutic purposes has become one of the modern frontiers of biomedical research. Thus, in silico modeling of macromolecular interactions has important motivations. My talk consists of two parts. First, I will describe the development of a physics based macromolecular docking method, which effectively uses geometry of the configuration space manifold. Specifically I will present application of generalized Fast Fourier Transforms on rotational group, for highly efficient global systematic search in the space of rigid body motions of one protein with respect to the other, and utilization of local optimization with exponential map parametrization of macromolecular interaction configurational manifold, for effective flexible protein refinement. I will demonstrate that the model is accurate enough not just to model the structure of the complex, but also provides biophysical insight in protein-protein association, which potentially enables mechanistic modeling of cellular crowding and therapeutics aggregation. The second part of the talk will focus on modeling protein small molecular interactions on the omics scale, building and expanding on ultra-fast algorithms described above. I will describe ongoing drug discovery and biological applications, including studying the kinome-wide possibility of targeting kinase allosteric sites with small molecules, and organism scale modeling of 3D structures of protein-metabolite interactions.
Dima Kozakov received a B.S. and M.S. in Applied Mathematics and Physics at the Moscow Institute of Physics and Technology, and PhD in Biomedical Engineering at Boston University. Currently he is Associate Professor in the Department of Applied Mathematics at Stony Brook University, and also Affiliate Member at Laufer Center for Physical and Quantitative Biology, Institute of Chemical Biology & Drug Discovery, and Institute of Advance Computational Sciences. Dr. Kozakov research focuses on the development of mathematically elegant, computationally efficient and physically accurate algorithms for modeling of biological macromolecules with emphasis on molecular interactions and drug design. Dr. Kozakov’s protein ligand docking approaches are consistently top performers in National Institute of Health (NIH) sponsored international competition D3R. His protein-protein docking method ClusPro has been consistently the best automatic server in the worldwide blind protein docking experiment CAPRI. Currently ClusPro has more than 20,000 users, with more than 200,000 jobs run in the last few years. The protein docking tools developed by Dr. Kozakov are licensed by Schrodinger, the largest pharmaceutical software vendor in the world, and are used by most major pharmaceutical companies worldwide. His research has been funded by the National Institute of Health, National Science Foundation, and Binational (US-Israel) Science Foundation.
The aortic valve interstitial cell (AVIC) is the most abundant cell type within the aortic valve and maintains the turnover of extracellular matrix (ECM) components. Under normal conditions, AVICs display a fibroblast-like, quiescent phenotype and can undergo myofibroblastic phenotypic activation in response to growth and disease which is characterized by an increase in overall cell contractility as well as an increase in ECM production and remodeling. Prolonged activation of AVICs can potentially cause drastic pathological changes in aortic valve ECM, geometry, and mechanical function and manifest into diseases such as aortic valve regurgitation or stenosis.
Cell contraction is a key component in biological processes such as wound-closure and can influence 3D tissue organization, remodeling, and function. Previous attempts to quantify cell contractility have largely relied on two dimensional assays such as traction force microscopy and micro-post assays which do not recapitulate the three dimensional complexity of native tissues. In this presentation, I will discuss our investigation of AVIC contractile behavior within 3D peptide-modified poly (ethylene glycol) (PEG) hydrogel matrices. We perform macro- and micro-level experiments to assess the contractile response of AVICs at the population- and single-cell levels, respectively. At the population-level, our results show that AVIC contraction within the PEG gel environment increases the overall stiffness of the AVIC-hydrogel construct and that AVIC contractile effects are highly dependent upon the adhesive ligand density within the PEG gel. In addition, we observed that the effects of AVIC contraction were more pronounced in lower stiffness hydrogels. At the single cell level, we used 3D traction force microscopy to assess AVIC contractile response and report that AVIC contraction is highly complex displaying contraction in one direction, expansion in an orthogonal direction, and virtually no deformation in the direction orthogonal to the previous two. In conjunction with our experimental investigations, we developed computational models of the macro- and micro-level experiments to gain a better mechanistic understanding of AVIC contractile behaviors. The macro-level model predicted that AVIC contraction causes an increase in AVIC-hydrogel construct stiffness and we observed that the Neo-Hookean material model was a good approximation of macro-level AVIC-hydrogel mechanical response. At the single-cell level, the Neo-Hookean material model is not sufficient to capture the effects of AVIC contraction on the hydrogel material, especially in regions further away from the cell surface. This discrepancy may have implications towards the complex and length-scale dependent AVIC-hydrogel interactions within our 3D culture system. The tunable and efficient techniques we have developed in this body of work will be used to quantify intrinsic differences in contractile properties between normal and diseased human AVICs to better our understanding of disease progression.
Alex Khang is a PhD candidate in the Department of Biomedical Engineering at The University of Texas at Austin. He received his BS degree in Biomedical Engineering from the University of Arkansas-Fayetteville. His doctoral project is focused on studying the mechanics and mechanobiology of heart valve interstitial cells (VIC) within a highly tunable hydrogel environment. He previously worked under the tutelage of Dr. Kartik Balachandran in the Mechanobiology and Soft Materials Laboratory. Fabricated nanoﬁbrous scaﬀolds for tissue engineering applications using a novel technique called centrifugal jet spinning.
Over the past decade, the BLAS-like Library Instantiation Software (BLIS) project has carefully revisited past progress on how to structure the implementation of the level-3 BLAS-like operations (matrix-matrix computations) in particular and all basic linear algebra operations in general. This has resulted in a refactoring of prior approaches that yields a more flexible, more easily maintained, highly portable, yet high-performing and scalable software library. BLIS now casts Goto's algorithm in terms of five portable loops (written in C99) around a “microkernel" that updates a small submatrix of C that fits in registers. It is only this microkernel that needs to be customized for a new architecture when implementing matrix multiplication. The refactoring exposed in BLIS drastically reduced the size, complexity, and number of assembly kernels necessary for supporting high-performance across all datatypes and level-3 operations. We will discuss how the science that underlies the high-performance implementation of GEMM, BLIS has opened up the field to new directions of research and to new contributors.
This work is in collaboration with current and former members of the Science of High-Performance Computing group and many external collaborators. It is/was supported by multiple grants from the National Science Foundation (including Awards ACI-1148125 and ACI-1550493) and gifts from AMD, Facebook, HP, Huawei, Intel, Microsoft, Oracle, Qualcomm, and Texas Instruments.
Robert van de Geijn is professor of computer science and member of the Institute for Computational Engineering and Sciences. He received his Ph.D. in Applied Mathematics from the University of Maryland, College Park. His interests are in linear algebra, high-performance computing, parallel computing, and formal derivation of algorithms. He heads the FLAME project, a collaboration between UT Austin, Universidad Jaume I (Spain), and RWTH Aachen University (Germany). This project pursues foundational research in the field of linear algebra libraries and has led to the development of the libflame library, a modern, high-performance dense linear algebra library that targets both sequential and parallel architectures. One of the benefits of this library lies with its impact on the teaching of numerical linear algebra, for which van de Geijn received the UT President’s Associates Teaching Excellence Award. He has published several books and more than 100 refereed publications.
The presentation will focus on a simulation and computational approach to verification of the hybrid mathematical models that are formed when combining physics-based models, with discrete-transition models such as those which model software algorithms. Namely, the mathematical models that arise when for instance considering Cyberphysical Systems, or the Internet-of-Things. In many game theory, filtering problems and verification problems it is not possible to analytically obtain solutions for statistical properties of systems under study. In the first section of the talk will concentrate on system verification, and will present a new verification algorithm for continuous-time stochastic hybrid systems, whose specifications are expressed in metric interval temporal logic (MITL), by deploying a novel model reduction method. By partitioning the state space of the hybrid system and computing the optimal transition rates between partitions, we provide a procedure to both reduce the system to a continuous-time Markov chain, and the associated specification formulas. We prove that the unreduced formulas hold (or do not) if the corresponding reduced formula on the Markov chain is robustly true (or false) under certain perturbations. In addition, a stochastic algorithm to complete the verification has been developed. We have extended the approach of this algorithm, and have developed a direct stochastic algorithm for probabilistic verifying a certain hybrid system class, and applied this technique to an extensive benchmark problem with realistic dynamics. In the second part of the talk we will describe our recent work on numerical approaches to obtaining estimates of statistical properties of Markov processes, in particular mean-square estimation. Monte Carlo simulation of Markov processes allows the numerical estimation of their statistical properties from an ensemble of sample system paths. We present methods for generating reduced-variance path ensembles for the tau-leaping discrete-time simulation algorithm, which allows mean stochastic process dynamics to be estimated with substantially smaller ensemble sizes. Our methods are based on antithetic and stratified sampling of Poisson random variates, and we provide a combination of analytical proofs and numerical evidence for their performance, which can frequently be a 2-3 orders of magnitude improvement over standard Monte Carlo. Also presented will be the HoTDeC multi-vehicle, which consists of indoor airborne and ground-based vehicles.
More than a hundred years ago Einstein, Smoluchowski, and Langevin formulated the laws of diffusion, and Perrin presented his systematic experiments tracking single, microscopic diffusing particles. Following several technological revolutions such as superresolution microscopy, experimentalists now measure the passive and active motion of submicron tracers and single molecules in complex systems such as living biological cells at unprecedented precision. Quite typically the measured motion significantly deviates from the laws of normal Brownian motion. Instead, anomalous diffusion is observed, in the form of non-Gaussian, long-range correlated, non-ergodic, or ageing dynamics. Based on state-of-the-art data from single particle tracking experiments and in silico systems this talk will elucidate the precise features unveiled in the data and established new theoretical approaches needed to understand the physical mechanisms behind the measured dynamics.
The main goal of my PhD work is to develop a robust, efficient, accurate, and stable finite element based framework to simulate small and large deformation solid mechanics’ problems involving complex geometries and complicated incompressible constitutive models. A third aspect, of practical importance, that is considered in my work is time-dependency. At the base of the aforementioned framework is the use of tetrahedral finite elements, with linear shape functions, and the variational multiscale approach.
The stability of the proposed algorithm was evaluated using an extensive set of numerical tests from the literature, as well as a thorough numerical analysis in the particular case of quasi-static J2-elasto-plasticity. From this extensive set of numerical tests, it was shown that the algorithm is invariant to the constitutive model and the time integrator used. Through a series of convergence tests, the algorithm was shown to have optimal rates of convergence, in the L2 norm, for the displacements, and velocities, and sub-optimal rate of convergence for the pressure. Finally, the robustness of the algorithm is showcased by considering realistic test cases involving complicated geometries and very large deformation.
Nabil Abboud is a Ph.D. candidate in the Department of Civil and Environmental Engineering (CEE) at Duke University. His research interests are Finite element, numerical analysis, fluid-structure interaction, computational solid mechanics, and computational fluid mechanics. His thesis, “Stabilized finite elements for problems involving complex constitutive models and irregular geometries” is in progress. He is advised by
Professor Guglielmo Scovazzi.
Disease of the heart valve are currently treated surgically with either repair or replacement strategies. Although repair strategies are preferred, they often fail prematurely due to the dynamic remodeling and signaling of living tissue. In order to understand the mechanisms of heart valve remodeling in disease scenarios, we use an integrated approach that leverages high content biological experiments with computational models of signal transduction. The ultimate goal of this work seeks to integrate these signal transduction models within a larger multiscale framework of specific heart valves to better understand the remodeling processes that occur in response to disease and predict optimal combinations of surgery and pharmaceutical interventions for different valve diseases.
Daniel P. Howsmon, PhD, is currently a postdoctoral fellow with the Willerson Center for Cardiovascular Modeling and Simulation at the Univ. of Texas at Austin. His research interests lie in using mathematical models and systems theory to shed light on complex biomedical problems and ultimately advance healthcare. He received his PhD in chemical and biological engineering from Rensselaer Polytechnic Institute in 2017 and BS degrees in both chemical engineering and biochemistry from Texas A&M Univ. in 2012.