A-Z Index | Phone Book | Careers

InTheLoop 11.30.2015

November 30, 2015

CRD's Demmel Chosen as AAAS Fellow

James Demmel of the Computational Research Division is among 347 new fellows named to the American Association for the Advancement of Science. Demmel, who holds a joint appointment with UC Berkeley, was cited for his distinguished contributions to the theory and practice of numerical linear algebra, especially for innovative approaches in parallel computing.

Election as a AAAS Fellow is an honor bestowed upon AAAS members by their peers in recognition of their achievements in advancing science or its applications. This year’s fellows were published in the Nov. 27 issue of the journal Science. They will be presented with an official certificate and a gold and blue rosette pin on Saturday, Feb. 13, at the 2016 AAAS Annual Meeting in Washington, D.C.

Founded in 1848, AAAS is the world’s largest general scientific society, and publisher of the journal Science. The association includes 254 affiliated societies and academies of science.

Five CS Staff to Receive Director’s Awards at Wednesday, Dec. 2 Ceremony

Five employees in the Computing Sciences organization will be recognized with Director’s Award for Exceptional Achievement at a ceremony to be held at 2:30 p.m. Wednesday, Dec. 2, in the Bldg. 50 auditorium.

Deb Agarwal, head of CRD’s Data Science and Technology Department, will be recognized for her effort to support diversity at the lab.

Eli Dart of ESnet’s Science Engagement Team and Brent Draney, head of NERSC’s Networking, Security and Servers Group, will be recognized for their work in developing the Science DMZ, a network architecture that allows scientists to exchange and access large data sets quickly and securely.

James Sethian, head of CRD’s Mathematics Group, will be recognized for establishing CAMERA, the Center for Advanced Mathematics for Energy Research Applications.

Lynn Rippe will be awarded a Berkeley Lab citation for her longtime procurement work in support of NERSC. »Read more.

Bashor to Co-chair New Disability Inclusion Employee Resource Group

Computing Sciences Communications Manager Jon Bashor and Marilyn Saarni of Earth and Environmental Sciences were chosen last week as co-chairs of the newly created Disability Inclusion Employee Resource Group. Kat Wentworth of the Director’s Office was named secretary and Micah Liedeker of the Office of the Chief Financial Officer is treasurer.

Organized under the lab’s Diversity and Inclusion Office, each employee resource group focuses on a primary diversity dimension that furthers Berkeley Lab’s business goals, in this case disability inclusion, sas reflected in this group's mission statement:

“The Berkeley Lab Disability Inclusion ERG advocates for a more accessible laboratory in design, in spirit and in operation. We work to create a welcoming and supportive environment for all employees and visitors and to serve as an information source for applicants, recruiters, managers and staff who have questions about disabilities, support and accommodations. In this way, we seek to increase understanding and appreciation for individuals with disabilities, whether visible or not. We see ourselves as part of a larger effort to build a more inclusive and respectful community, knowing that when improvements are made to help one group, everyone benefits.”

This Week's CS Seminars

Applied Math Seminar: Modeling reactive events in complex systems

Tuesday, Dec. 1, 3–4pm, Bldg. 50B, Room 4205
Thuan Nguyen, Oregon Health & Science University Portland State University (OHSU/PSU) School of Public Health

Fence method is a recently proposed strategy for model selection. It was motivated by the limitation of the traditional information criteria in selecting parsimonious models in some nonconventional situations, such as mixed model selection. Jiang and others adaptively simplified the method to make it more suitable and convenient to use in a wide variety of problems. Still, the current modification encounters computational difficulties when applied to high-dimensional and complex problems. To address this concern, we proposed a restricted fence procedure that combines the idea of the fence with that of the restricted maximum likelihood. Furthermore, we propose to use the wild bootstrap for choosing adaptively the tuning parameter used in the restricted fence. We focus on problems of longitudinal studies and demonstrate the performance of the new procedure and its comparison with other procedures of variable selection, including the information criteria and shrinkage methods, in simulation studies. The method is further illustrated by an example of real-data analysis. This work is joint with Jiming Jiang, University of California, Davis

Applied Math Seminar: Modeling reactive events in complex systems

Wednesday, Dec. 2, 3:30Y–4:30pm, 939 Evans Hall, UC Berkeley
Eric Vanden, Eijnden

Reactive events such as conformation change of macromolecules, chemical reactions in solution, nucleation events during phase transitions, thermally induced magnetization reversal in micromagnets, etc. pose challenges both for computations and modeling. At the simplest level, these events can be characterized as the hopping over a free energy barrier associated with the motion of the system along some reaction coordinate. Indeed this is the picture underlying classical tools such as transition state theory or Kramers reaction rate theory, and it has been successful to explain reactive events in a wide variety of context. However this picture presupposes that we know or can guess beforehand what the reaction coordinate of the event is. In many systems of interest – protein folding, enzyme kinetics, protein-protein interactions, etc. – making such educated guesses is hard if not impossible. The question then arises whether we can develop a more general framework to describe reactive events, elucidate their pathway and mechanism, and give a precise meaning to a concept such as the reaction coordinate. In this talk I will discuss such a framework, termed transition path theory (TPT), and indicate how it can be used to develop efficient algorithms to accelerate the calculations and analysis of reactive events. I will also illustrate these concepts on several examples, including the reorganization of Lennard-Jones clusters and the folding of the pinWW domain protein.

High resolution whole brain histology to MRI registration: applications and new challenges

Thursday, Dec. 3, 12-1pm, Bldg. 50F, Room 1647
Maryana Alegro, PhD, University of California, San Francisco

Medical Imaging revolutionized the field of neuroscience by providing tools for examining the living brain, with MRI being the most popular by far. Despite technological advances, clinical brain MRI use is rather limited since its resolution is still 1,000x lower than histology. We developed a pioneering computational framework to co-register brain MRI to its histology counterpart using postmortem brain. It is already in use in projects sponsored by governmental and non-governmental entities. However, the pipeline is limited to low-resolution images, since registration of higher-resolution histology incurs in a larger computational problem. Our goal is to improve our pipeline and develop algorithms to register high-resolution histology to MRI, in order to allow better correlation between MRI signal and histological features. MRI signals validated by their corresponding histology would be incorporated in neuroscience studies to suggest the most probable brain changes associated with each specific MRI signal feature. The impact will be a broader clinical application of brain MRI in the clinical practice, e.g., by accelerating biomarkers discovery and enabling clinical trials, in addition to enabling experts to browse through the high-resolution whole brain volumes through different layers of information, captured from combined imaging modalities.

BIDS Data Science Lecture:Identification of Novel Cell Types Using Single-Cell Transcriptome Sequencing

Friday, Dec. 4, 1–2:30pm, 190 Doe Library, UC Berkeley
Sandrine Dudoit, UC Berkeley

Single-cell transcriptome sequencing (scRNA-Seq), which combines high-throughput single-cell extraction and sequencing capabilities, enables the transcriptome of large numbers of individual cells to be assayed efficiently. Profiling of gene expression at the single-cell level is crucial for addressing many biologically relevant questions, such as the investigation of rare cell types or primary cells (e.g., early development, where each of a small number of cells may have a distinct function) and the identification of subpopulations of cells from a larger heterogeneous population (e.g., discovering cell types in brain tissues). scRNA-Seq assays generate large datasets and involve inference for high-dimensional multivariate distributions with complex and unknown dependence structures among variables. Dudoit will discuss some of the statistical analysis issues that have arisen in the context of a collaboration funded by the Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative, with the aim of classifying neuronal cells in the mouse somatosensory cortex. These issues, ranging from so-called low-level to high-level analyses, include exploratory data analysis (EDA) for quality assessment/control (QA/QC) of scRNA-Seq reads, normalization to account for nuisance technical effects, cluster analysis to identify novel cell types, and differential expression analysis to derive gene expression signatures for the cell types.