Three EPSRC funded PhD studentships are available at the University of
Warwick via the Warwick Centre for Analytical Science (http://www2.warwick.ac.uk/fac/sci/wcas/ <https://mywebmail.warwick.ac.uk/exchweb/bin/redir.asp?URL=http://www2.warwick.ac.uk/fac/sci/wcas/> ).
These projects aim at reinforcing the role of statistical
modeling in chemistry and physics, focusing on the integration of mass
spectrometry and electron microscopy data in space and time.
Informal enquiries (either about these specific projects and about potential interest in this area)
should be made to Dr. Fabio Rigat ([log in to unmask])
Applications must be made via the University website at the URL:
http://www.go.warwick.ac.uk/pgapply <https://mywebmail.warwick.ac.uk/exchweb/bin/redir.asp?URL=http://www.go.warwick.ac.uk/pgapply>
PROJECT 1: Statistical Methods in Mass Spectrometry
This is a collaborative project between the group of Professor Peter
O'Connor (Chemistry) and that of Dr. Fabio Rigat (Statistics) at Warwick.
Modern mass spectrometers are equipped with standard data analysis
packages which perform such tasks as peak-picking, isotopic envelope
deconvolution, charge state determination, etc. However, commercial
methods are typically not robust, working well for intense, isolated
peaks and very poorly for low intensity, overlapping peak
distributions. The net result is that information content in the spectra is lost or ignored.
The goal of this collaborative project is to develop an integrated approach to modeling mass spectrometry experiments. Mathematical models of the raw time-domain signal will be combined
with appropriate statistical methods for learning signal features from
noisy experimental data and for reflecting the associated uncertainty
on spectral estimates. Bayesian inference methods will be used to
provide a formal probabilistic link across related mass spectrometry
experiments, for instance leading to a dynamic characterization of
post-translational protein modifications and to the integration of mass spectrometry data for different proteins.
PROJECT 2: Use of statistics and trajectory modeling to improve mass accuracy in Fourier-transform mass spectrometry
This is a collaborative project between the group of Professor Peter
O'Connor (Chemistry) and that of Dr. Fabio Rigat (Statistics) at Warwick.
This project aims at integrating modern ion
trajectory modeling with experimental data to extract and correct for
"space-charge" perturbations to the ion frequencies of motion in an
electric/magnetic field. "Space-charge" refers to the multiple
coulombic repulsions present within the ion trap during detection of
the mass spectra. These coulombic repulsions shift the detected
frequencies in a systematic way, but because the ions are constantly
moving, the magnitudes of the shift are constantly changing.
This space-charge error term is the largest for mass determination
and methods to fully correct for it could result in a significant
improvement in mass accuracy.
The project will involve developing multivariate models to the raw
time-domain mass spectrum data. This development will include existing
univariate ion trajectory models for detecting relevant ion signals
and flexible correlation structures fitting the space-charge
interaction terms.
PROJECT 3: Joint Modeling of Electron Microscopy and Mass Spectrometry data for
protein structural characterization.
This is a collaborative project between the group of Professor Peter
O'Connor (Chemistry), Dr. Jeremy Sloan (Physics) and that of Dr. Fabio Rigat (Statistics) at Warwick.
Modern Mass Spectrometry (MS) methods can determine exactly which
amino acid in a protein is modified, and by exactly what moiety (with
some limitations). Modern Electron Microscopy (EM) can visualize, at
a sub-Angstrom level (now typically in the range 0.7-0.8 Å spatial
resolution), the positions of individual atoms in a molecule,
particularly if heavy nuclei are incorporated.
This project aims at developing statistical models for the joint
analysis of MS and EM data. A key goal of this project is that of improving the accuracy of the
EM structural reconstructions by modeling the spatial relationships
among protein components which masses are measured via MS. In his
context, spatial regression models inspired by the physics and
chemistry of the proteins being measured are instrumental to
discriminate significant relationships from noise.
|