2012-2013

 

April 26, 2013

Evolution of the Cancer Genome

Speaker: Franziska Michor, Associate Professor, Biostatistics and Computational Biology, Dana-Farber Cancer Institute and Harvard School of Public Health

Cancer emerges through an evolutionary process in somatic tissue. The fundamental laws of evolution can best be formulated as exact mathematical equations; therefore, the process of cancer initiation and progression is amenable to mathematical investigation. Current areas of research of the Michor lab include cancer stem cells, the evolution of drug resistance, and the dynamics of metastasis formation. In this talk, I will introduce two examples of the application of evolutionary theory to cancer genomics and treatment.

 

April 12, 2013

Designing Visualizations for Biological Research

Speaker: Miriah Meyer, USTAR Assistant Professor in the School of Computing and faculty member in the Scientific Computing and Imaging Institute, University of Utah

Advances in measurement devices in the last decade have given rise to an explosion of scientific data--data that hold the promise of curing human disease, predicting the future health of our planet, and unlocking the secrets of the universe. In biology, access to massive amounts of quantitative data has fundamentally changed how discoveries are made. Today, making sense of this data using visualization methods is an important component of the scientific process. The visualization toolbox of most biologists, however, contains nothing to supplement broadly available tools that were designed for overarching problems and that often leave them without answers to their specific research questions. A growing trend in the visualization community is to develop tools that focus on specific, real-world problems. Called a design study, the process of developing these tools relies on a close collaboration with end users as well as the use of methods from design. In this talk I'll present several design studies that target complex, biological data analysis, from discovering trends in molecular networks to understanding the results of comparative genomics algorithms.

 

March 29, 2013

Possible Infrastrucures for Data Science

Speaker: Jeff Hammerbacher, Founder and Chief Scientist, Cloudera; Assistant Professor, Mount Sinai School of Medicine

We'll explore three potential infrastructures for data science and speculate on what new infrastructures may emerge in the future.

 

March 15, 2013

Multi-scale Seizure Dynamics

Speaker: Mark Kramer, Assistant Professor of Mathematics, Boston University

Epilepsy—recurrent, unprovoked seizures—is a common brain disease, affecting 1% of the world’s population. Seizures are typically identified as abnormal patterns in brain voltage activity. Many open questions surround epilepsy and seizures, and identifying the answers promises new insights for treatment and prevention. In this talk, I will consider brain voltage activity during seizures as observed at multiple spatial scales. I will show how techniques from mathematics and statistics can be used to characterize these data, identify common features, and connect observed brain activity to mechanisms. One specific open question is this: Why do seizures spontaneously terminate? Analysis of human brain electrical activity at various spatial scales suggests a common dynamical mechanism: a discontinuous critical transition or bifurcation. Prolonged seizures (status epilepticus) repeatedly approach, but do not cross, the critical transition. I will consider a computational model to demonstrate that alternative stable attractors, representing the seizure and post-seizure states, emulate the observed brain dynamics. These results also suggest a dynamical understanding of status epilepticus. Seizure dynamics thus provide an accessible system for studying critical transitions in nature.

 

March 1, 2013

The Network Effect: Integrative Systems Approaches to Modeling Biological Processes

Speaker: John Quackenbush, Professor of Biostatistics and Computational Biology and Professor of Cancer Biology, Dana-Farber Cancer Institute; Professor of Computational Biology and Bioinformatics, Harvard School of Public Health

Two trends are driving innovation and discovery in biological sciences: technologies that allow holistic surveys of genes, proteins, and metabolites, and the growing realization that analysis and interpretation of the results requires an understanding of the complex factors that mediate the link between genotype and phenotype. The growing body of biological and biomedical information, driven by an exponential drop in the cost of generating genomic data, provides an outstanding opportunity for leveraging what we already “know” in a systematic way to understand the problems we are studying. Here, I will provide an overview of some of the methods we are using to investigate the complexities of human phenotypes and to explore how we can use biological data to uncover the cellular networks and pathways that underlie human disease, building predictive models of those networks that may help to direct therapies.

 

February 22, 2013

Multiphysics and multiscale modeling of cardiac dynamics

Speaker: Boyce Griffith, Assistant Professor of Medicine; Affiliate Faculty Member of the Department of Mathematics, Courant Institute of Mathematical Sciences, New York University

The heart is a coupled electro-fluid-mechanical system. In this talk I will present mathematical models and adaptive numerical methods for describing cardiac mechanics, fluid dynamics, and electrophysiology, as well as applications of these models and methods to cardiac fluid-structure and electro-mechanical interaction. I will also describe novel models that go beyond traditional macroscopic descriptions of cardiac electrical impulse propagation by explicitly incorporating details of the cellular microstructure into the model equations. Standard models of cardiac electrophysiology account for this cellular microstructure in only a homogenized or averaged sense, and we have demonstrated that such homogenized models yield incorrect results in certain pathophysiological parameter regimes. To obtain accurate model predictions in these parameter regimes without resorting to a fully cellular model, we have developed an adaptive multiscale model of cardiac conduction that locally deploys detailed cellular models only where needed, while employing the more efficient macroscale equations where those suffice. I will present applications of these methods to whole heart models, detailed models of cardiac valves and novel models of aortic dissection.

 

February 15, 2013

The Giza Pyramids, 3D Technologies, and the Challenges of Archaeological Information Management

Speaker: Peter Der Manuelian, Philip J. King Professor of Egyptology, Harvard

As part of the Giza Project at Harvard, a 3D, archaeologically accurate computer model of the pyramids, tombs, and temples at the famous Giza Pyramids, just west of modern Cairo, is being used for teaching and research. The work is largely based on the excavations of the Harvard University–Boston Museum of Fine Arts Expedition (1905–47). This talk will show aspects of the computer model and present other experiments in new technologies for bringing the site back to life, for scholars, students, and the public worldwide.

 

February 13, 2013

Exploring Energy Landscapes: From Molecules to Nanodevices

Speaker: David Wales, University Professor of Chemical Physics, University of Cambridge

In molecular science, a computational framework for investigating structure, dynamics and thermodynamics can be provided by coarse-graining a potential energy surface into the basins of attraction of local minima. Steps between local minima form the basis for global optimization and for calculating thermodynamic properties. To treat global dynamics, we must include transition states of the potential energy surface. These link local minima via steepest-descent paths. We may then apply discrete path sampling, which provides access to rate constants for rare events. In large systems the paths between minima with unrelated structures may involve hundreds of stationary points of the potential energy surface. New algorithms have been developed for both geometry optimization and finding connections between distant local minima. Applications will be presented for a range of different examples, including atomic and molecular clusters, biomolecules, condensed matter, and coarse-grained models of mesoscopic structures.

 

November 30, 2012

Will "Big" Data Yield Big Insights about Human Society?

Speaker: David Lazer, Professor of Political Science and of Computer and Information Science, Northeastern University; Visiting Research Fellow, Harvard

The emergence of "big data" about human behavior combined with ever-escalating computational power offers the possibility of paradigm-shattering research about society. Human behavior is being captured in minute detail as never before--from our political opinions expressed in various social media, to our various personal and professional communications expressed in various electronic forms, to our most intimate questions, as captured by Google. These massive, passive data, however, are not designed for scientific purposes. The objective of this talk is to discuss the scientific opportunities and pitfalls these various data sources offer, as well as to examine institutional barriers to the development of a computational social science.

 

November 16, 2012

Design and Analysis of Experiments that Leverage Social Structure and Interactions

Speaker: Edoardo Airoldi, Assistant Professor of Statistics, Harvard

A number of scientific endeavors of current national and international interest involve populations with interacting and/or interfering units. In these problems, a collection of partial measurements about patterns of interaction and interference (e.g., social structure and familial relations) is available, in addition to more traditional measurements about unit-level outcomes and covariates. Formal statistical models for the analysis of this type of data have emerged as a major topic of interest in diverse areas of study. In this talk, I will review a few ideas and open areas of research that are central to this burgeoning literature, placing emphasis on inference and other core statistical issues. Then I will turn to describing a technical notion of non-ignorability that applies to sampling designs that leverage social structure, an inference strategy that can be used to obtain valid estimates in these settings, and a randomization-based approach to estimating the causal effect of peer influence, with hints to applications to advertising on social media platforms, politics and healthcare in which these statistical problems arise.

 

November 2, 2012

A Computational Engineer Combusts

Speaker: Margot Gerritsen, Associate Professor, Energy Resources and Engineering and Director, Institute for Computational and Mathematical Engineering, Stanford University

Large-scale production of very heavy oil is gaining momentum. Unfortunately, production of such reservoirs typically leads to large environmental impacts. One promising technique that may mitigate these impacts is in-situ combustion (ISC). In this process, (enriched) air is injected into the reservoir. After ignition a combustion front develops in situ that burns a small percentage of the oil in place and slowly moves through the reservoir producing steam along the way. The steam moves ahead of the front, heats up the oil, makes it runnier and hence easier to produce. A side benefit of this process is that the heat thus generated often cracks the oil into heavy, undesirable components that stay behind in the reservoir and lighter, more valuable components that can be brought up to the surface. In the last few years, my colleagues and I plunged into heavy oil recovery to see if computational mathematics could make a difference in pushing this process over less environmentally friendly processes in the industry. ISC processes are notoriously hard to predict. We developed a workflow involving laboratory experiments, various simulation tools and upscaling methods that increases the confidence of the oil reservoir engineer in ISC. We hope that this will lead to a wider acceptance and use of this technique.

 

October 19, 2012

Overview of Scientific Activities at D. E. Shaw Research

Speakers: Paul Maragakis and Zhou Fan, Research Scientist and Scientific Programmer, D. E. Shaw Research

We at D. E. Shaw Research (DESRES) are engaged in computational biochemistry with a long-term goal of impacting biomedical research. Our group has built a specialized supercomputer, Anton, that has enabled millisecond-long molecular dynamics simulations of proteins. This talk will provide an overview of our group and will review several of our recent, high-impact scientific results. The speakers will also answer questions about working at DESRES.

 

October 5, 2012

Leveraging High-Performance Parallel Computing for Biologically Inspired Object and Face Recognition

Speaker: David Cox, Assistant Professor of Molecular and Cellular Biology and the Center for Brain Science, Harvard

The visual cortex of the human brain is unrivaled by artificial systems in its ability to recognize faces and objects in highly variable and cluttered real-world environments. Biologically inspired computer vision systems seek to capture key aspects of the computational architecture of the brain. Such approaches have proven successful across a range of standard object- and face-recognition tasks. However, since the number of parameters in a vision model is typically large, and the computational cost of evaluating one particular parameter set is high, when a model fails we are left uncertain whether it is because we are missing a fundamental idea, or because the correct "parts" have not been tuned correctly, assembled at sufficient scale, or provided with enough training. In this talk, I'll present a high-throughput search approach for exploring a broad range of biologically inspired vision models. I'll discuss parallel programming techniques that enable this approach, including machine-learning-guided metaprogramming techniques that bring the ideas of high-throughput search down to the level of implementation-level code optimization.

 

September 21, 2012

Scientific Computing for Movie Special Effects and Virtual Surgery

Speaker: Joseph Teran, Associate Professor of Applied Mathematics, UC Los Angeles

I will talk about some exciting new applications of scientific computing for solid and fluid mechanics problems, including the simulation of virtual materials for movie special effects, video game effects and virtual surgery. These new applications all have an increasing demand for physically realistic dynamics of materials like water, smoke, fire, brittle objects, elastic objects, etc. Their computational demands are somewhat different than those traditionally considered by scientific computing researchers, and many new algorithms are needed to address them. A virtual surgery simulator, for example, is like a flight simulator for training surgeons (and would-be surgeons) in modern procedures. I will discuss procedures related to repair and manipulation of soft tissues. Other topics discussed will include GPU and many-core algorithms for real-time solution of nonlinear elliptic equations arising in elasticity problems and in incompressible flow, cut-cell methods for higher-order accuracy on structured grids, and contact algorithms for thin structures. I will discuss these challenges as well as some recent algorithms developed in my lab to address them.

 

September 7, 2012

Efficient Simulation of Multiscale Kinetic Transport

Speaker: Nicolas Hadjiconstantinou, Professor of Mechanical Engineering and Director of Computation for Design and Optimization, MIT

This talk will discuss a new class of approaches for simulating multiscale kinetic problems, with particular emphasis on applications related to small-scale transport. These approaches are based on an algebraic decomposition of the distribution function into an equilibrium part, described deterministically (analytically or numerically), and the remainder, described using a particle simulation method. The discussion will pay particular attention to stochastic particle simulation methods that are typically used to simulate kinetic phenomena. Algebraic decomposition can be thought of as control-variate variance-reduction formulation, with the nearby equilibrium serving as the control. Such formulations can provide substantial computational benefits; in many cases, the computational cost reduction is sufficiently large to enable otherwise intractable simulations. The proposed methods will be illustrated with a variety of problems of engineering interest, such as microscale/nanoscale gas flows.