Seminars

In biological and material sciences, one is often interested in the three-dimensional structures of cells, molecules and solids. The two main approaches scientists at the Berkeley lab use to obtain microscopic structure information are imaging and first principle based simulation. In the first approach, various types of light sources and imaging devices are used to produce partial views of the object of interest. These partial views are then combined and analyzed computationally to provide a more detailed characterization of the object at a desired resolution. In the second approach, structure properties are deduced from analyzing the interactions among the basic constituents of the object . In this talk, I will describe some of the computational problems that arise from these approaches, how they are currently tackled and the challenges ahead. In particular, I will focus on the problem of three-dimensional protein structure determination using cryo-electron microscopy data and electronic structure calculations for molecules and solids.

A detailed understanding of how organisms respond to environmental stress requires knowledge of not only the genetics and physiology of stress responses, but also their dynamics and mechanisms of evolution. Humans, animals, plants, and microbes live in constantly changing environments under frequently adverse conditions. To survive, they must have sophisticated stress responses. These stress responses are carried out by complex gene regulatory networks that accurately sense the stress, choose the appropriate response (whether fight, flight, or non-violent resistance), and carry it out. However, because of the changing environment, the stress responses must also retain evolvability, i.e. the ability to cope with previously-unseen circumstances. Understanding this tradeoff between complexity and evolvability is a critical aspect of the challenge before us. In this talk, I will describe preliminary progress in characterizing the evolution of bacterial stress responses, which play a particularly important role in human disease and the biodegradation of toxic wastes.

Electrostatic interactions play an important role in many biological
problems and can lead to counterintuitive phenomena. I will highlight a
number of problems in this area that we have addressed by means of
computational methods. Specifically, we have used Monte Carlo and
molecular dynamics simulations to better understand the self-assembly of
stiff polyelectrolytes (charged polymers). Such molecules,
e.g. filamentous actin, form close-packed bundles in the presence of
multivalent ions or proteins. We elucidate the mechanism of this
self-assembly process and are able to make direct comparison to
experimental results obtained via small-angle x-ray scattering. I will
also demonstrate how these findings pertain to fighting bacterial
infections in cystic fibrosis patients.

Abstract: Applications using automatically indexed clinical conditions must account for contextual information such as whether a condition is negated, historical, hypothetical, or experienced by someone other than the patient. We developed and evaluated an algorithm called ConText, an extension of the NegEx negation algorithm, which relies on trigger terms, pseudo-trigger terms, and termination terms for identifying the values of these contextual properties. We will describe an evaluation of ConText’s performance on Emergency Department reports, an evaluation of ConText’s portability to five other report types, and current work motivated by ConText's errors.

Bio: Dr. Wendy Chapman is an assistant professor in the Department of Biomedical Informatics at the University of Pittsburgh. She received a B.A. in Linguistics and a Ph.D. in Medical Informatics from the University of Utah. Dr. Chapman’s main interest is applying natural language processing (NLP) techniques to clinical records. The majority of her research has focused on information extraction from dictated clinical records, including chest radiograph reports, Emergency Department reports, and chief complaints. Her research has addressed generating reliable annotations from manual annotators, identifying clinical conditions that are suggestive of particular diseases or syndromes, and identifying modifying information important for understanding a patient’s clinical state, including negation, uncertainty, and temporal information. Her work has been applied to decision support and biosurveillance.

Realistic Computer Simulation of Cell Signaling Using MCell: Evidence
for Ectopic Neurotransmission at a Neuronal Synapse


Biophysically accurate computational experiments performed on cell signaling
pathways are a powerful way to study cell signaling, helping to formulate and
test new hypotheses in conjunction with bench experiments. MCell is a computer
program that allows realistic 3D simulations of biochemical signaling pathways
in and around complex subcellular ultrastructure. Here I will introduce
fundamental concepts of cell signaling processes in organized, compact spaces
and discuss the computational methods employed by MCell to simulate these
processes. I will then demonstrate the use of MCell to explore synaptic
transmission at a unique type of synapse located in the chick ciliary ganglion
(CCG). Neurotransmitter release is well known to occur at specialized synaptic
regions that include presynaptic active zones and postsynaptic densities. At
cholinergic synapses in the CCG, release of acetylcholine activates two
distinct types of nicotinic acetylcholine receptors (nAChRs): alpha-3 nAChRs
located primarily at active zones, and extrasynaptic alpha-7 nAChRs located on
specialized membrane formations. Physiological measurements suggest that
release distant from postsynaptic densities can activate the predominantly
extrasynaptic alpha-7 nAChR subtype. We used MCell to explored such ectopic
neurotransmission in a realistic computer model of a CCG synapse reconstructed
via high-resolution serial electron microscopic tomography. Simulated
synaptic activity is consistent with experimental recordings of miniature
excitatory postsynaptic currents only when ectopic transmission is included in
the model. I will present this work as well as new experimental findings by
our collaborator, Peter Sargent, which have now largely confirmed this
prediction and have uncovered new surprises leading to even more intrigue in
this unfolding story.

Integrated Genomics Approaches to Discovering the Genetic
Basis of Complex Traits in Inbred Mouse Strains

Inbred mouse strains are a very powerful and well studied human
disease and complex trait model. A tremendous amount of information
is available for various inbred strains including phenotypic
information stored in the Mouse Phenome Database (MPD) and
high-throughput genomic data such as expression microarray data.
Recently, several high density SNP maps have also been developed for
inbred mouse stains. These resources combined with what is already
known about mouse genetics in terms of quantitative trait loci (QTLs)
and known pathways, make inbred mouse strains an ideal model system.
We present results of our analyses which combines multiple types of
data in order to understand the genetic basis of complex traits. We
perform whole genome association analysis of the mouse SNP maps over
the phenotypes in the MPD. We describe how we augment our association
analysis results with information from expression data, known pathways
and QTLs. We demonstrate the how our approach is able to discover
many regions in the mouse genome associated with phenotypes and how
many of our predictions are consistent with genes known to influence
specific traits.

The dynamics of a complex system depends on the topological structure of how its pieces are connected and how those pieces interact with each other. The interesting dynamics of such systems will be illustrated with two examples. Genes can express transcription factor proteins that regulate the expression of other genes. The levels of gene expression, as determined by the mRNA levels measured by microarrays, can provide information on the topology of the gene regulatory network. People influence the emotional or behavioral state of other people. Dynamical models of these interactions can match previous observations and make new predictions that can be verified, or contradicted, by social psychology experiments. It's interesting that the feedback functions chosen to represent human social interactions can also shed light on how we should think about the biochemical interactions between genes. (References: http://www.ccs.fau.edu/~liebovitch/genes.pdf , http://ssrn.com/author=913360 ).

Alternative splicing of precursor mRNAs is a prevalent mechanism of gene
regulation in higher eukaryotes. It generates enormous transcriptome
diversity from a limited repertoire of protein-coding genes in the genome.
Almost three quarters of multi-exon genes in the human genome are
alternatively spliced. In this talk, I will describe the use of high-density
exon tiling microarrays for global analysis of pre-mRNA alternative
splicing. I will introduce a new computational method for analysis of
alternative splicing using Affymetrix Exon 1.0 array data. By recognizing
and correcting for microarray noise such as background and
cross-hybridization, our method can identify differential alternative
splicing events at a very low false positive rate. I¹ll also discuss the
impact of alternative splicing on the evolution of mammalian genomes,
inferred from comparative analyses of transcriptome data from multiple
species.

The application of software development methods for scientific and
biological research tools has been considered a desirable approach.
Through a number of examples, we illustrate the advantages of applying
such methods in terms of software development, maintenance, and addition
of new functionality. Specifically, we describe our current experience
with Predictive Health Initiative, a collaboration between Emory
University and Georgia Tech, to explain the application of modern
software development methods for data and system integration. For
example, we use domain-specific code generation to integrate data from
heterogeneous and evolving data sources. In addition, we apply concepts
and techniques from event processing and services computing to implement
required functionality that would have been prohibitively expensive
without them

The standard diffraction limit of light is about 250 nm, meaning that you cannot "resolve" objects closer than this distance. Despite this, we have come up with a method to measure 1.5 nm in x-y plane, with 1-500 msec, using a technique we call Fluorescence Imaging with One Nanometer Accuracy (FIONA). We have chosen to study molecular motors, both in vitro and in vivo. We find that all tested motors walk in a hand-over-hand fashion, in vitro and in vivo. We also find evidence that in vivo, two of the same motors carry cargo simultaneously-but not cooperatively. We also see passing of cargo from one type of motor to another. Finally, we find that, in vivo, things walk around a microtubule, not just co-linear with them.