\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\big DEPARTMENTAL SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{4:15 p.m., Tuesday, March 2, 2004}
\centerline{Sequoia Hall Room 200}
\centerline{(Cookies at 3:45 in 1st Floor Lounge)}

\bigskip
\baselineskip=15pt
\centerline{\sl Samuel Kou}
\centerline{\sl Harvard University}
\bigskip
\centerline{\bf Statistical Analysis of Single Molecule Experimental Data}
\bigskip

Abstract:

Recent advances in experimental technologies allow scientists to follow
biochemical processes on a single molecule basis, which provide much
richer information about chemical dynamics than traditional ensemble
averaged experiments, but also raise many new statistical challenges. This
article provides the first likelihood-based statistical analysis of the
single-molecule fluorescence lifetime experiment designed to probe the
conformational dynamics of a single DNA hairpin molecule. The conformational
change is initially modeled as a continuous-time two-state Markov chain,
which has to be indirectly inferred from changes in photon emissions on
top of the unobservable molecular Brownian diffusion. Beyond the simple
two-state model, a competing model that models the energy barrier between
the states of the DNA hairpin as an Ornstein-Uhlenbeck process has been
suggested in the literature. We first derive the likelihood function of
the simple two-state model using a matrix discretization method, and then
generalize the method to handle complications such as unobserved molecular
diffusions and the fluctuating energy barrier. The data augmentation
technique and Markov chain Monte Carlo methods are developed to sample
from the desired posterior distribution. The Bayes factor calculation
together with posterior estimates of relevant parameters indicate that the
fluctuating barrier model fits the data better than the simple two-state
model.

This work is joint with Sunney Xie and Jun Liu


\bye