Department of Computer Science
Computational Science and Engineering Group
My office is in Harold Frank Hall, Room 5110.
My mailing address is:
Computer Science DepartmentResearch
University of California Santa Barbara
Santa Barbara, CA 93106-5110
USA
I work on problems within the broad field of Systems Biology. My current research is divided equally between two topics:
My thesis work addresses the impact of space on the Stochastic Simulation
Algorithm. The SSA was originally derived in the point molecule limit, and
its correct application hinges on the system being "well-stirred" (i.e. the positions of the
molecules being uniformly random). I study the impact of relaxing each of
these assumptions.
In the biological setting, large molecules or many molecules are often crowded
into small spaces. To begin to address the impact of reactant-excluded volume,
we've looked at a simple A+A reaction in a ballistic setting in one and two
dimensions. We've shown analytically, and confirmed computationally, that in
one dimension, the reactant-excluded volume has a predictable impact on the
propensities of the SSA (Gillespie, Lampoudi, Petzold, 2007).
In two dimensions it is impossible to study this question
analytically, but I've developed two methodologies (one exact and one
approximate) which have allowed us to study it computationally. We've found
that at large molecule populations the effect of reactant-excluded
area is close to what one might expect, but at small molecule populations
reactants exclude larger than expected area (Lampoudi, Gillespie, Petzold,
submitted).
The "well-stirred" assumption often doesn't hold true for systems which
we want to simulate stochastically. The Inhomogeneous Stochastic Simulation
Algorithm (ISSA) is the spatially dependent variant of the SSA customarily
used to address that setting.
While the
algorithm per se is straightforward, its physical underpinnings, range
of validity and method of application to any given biochemical system
remain to be elucidated. In particular my research addresses what
spatial resolution can be expected to be obtainable for any given
model system, as well as how the physics underlying the algorithm
could be clarified. The ISSA is an extremely time consuming algorithm,
since it simulates every single reaction and diffusion event.
I am also looking at hierarchies of more efficient
approximations to the exact algorithm.
The second project in which I am engaged involves the development of a computational model of a bacterial phosphorelay signal transduction cascade. I am collaborating with the Cotter group (Dept of Molecular, Cellular and Developmental Biology, UCSB) in an attempt to build a model of the BvgAS His-Asp-His-Asp phosphorelay and transcriptional regulation system, which is believed to control virulence in the Bordetella family of pathogenic bacteria. This is a methodical process, in which the final intended uses of the model -- answering questions of biological interest -- must always be kept in mind. Among the methods used are the numerical solution of ordinary differential equations, numerical optimization, model reduction and ultimately multiscale simulation of the resulting models.
In the past I have worked on molecular dynamics simulations of protein folding as well as problems from the more traditional field of high performance computing. The best way to find out about those endeavors is to look at my publications.
Journal and Refereed Publications
Curriculum Vitae (PDF)
SWILL (Simple Web Interface Link Library). The paper won the "best student paper" award at Freenix '02 (ps) (html). SWILL was also named "NetBSD package of the month" for July 2002. Thanks!
On the left, keeping a lookout on Nevermind on the way from Monterey to SB. On the right, hacking on Shadow's masthead.
I spend my free time sailing my boat, the Shadow Line. More about that here.