Selection Bias in Multiple Regression
When multiple linear regression models are constructed from a large pool of potential independent variables they suffer from an effect known as selection bias, which has the effect of making the resulting models appear more significant than they really are. This phenomenon is being investigated both analytically and using Monte Carlo simulations.
Parasurf Project
A novel QM, non atom based methodology for virtual screening based on molecular surface properties.
Previous projects
Topological Descriptions of Molecular Shape
Topological invariants of molecular shape based on van der Waals surface graphs are being studied for use in molecular similarity and QSAR.
Zeolite Modelling
Molecular modelling methods are being applied to inorganic materials to predict and rationalise structures and reaction mechanisms.
Biological and Chemical Data Mining
Extraction of decision trees from artificial neural networks trained on bioinformatic and chemoinformatic data.
Mode of Action of Destruxin
Studies of the mode of action of destruxin, a cyclic hexadepsipeptide, in collaboration with the BBSRC IACR Rothamsted.
Non-Structural Database Searching
The aim of this work is to create an environment in which compounds can be ranked by similarity to a target using not only the structural data, but also other data associated with the compounds (e.g. boiling point, toxicity, etc.).
Drug-DNA Interactions
Molecular mechanics and dynamics methods can be used to rationalise and predict the interactions between drugs and DNA.
Improved Mathematical Methods for Drug Design
Multivariate statistical techniques can be used to model the biological activity of drugs and agrochemicals with the aim of producing robust and accurate predictions.