GENERAL APPLICATIONS SECTION
Use of statistical modelling in chemistry
Tuesday 8th October 2002, 5pm (Tea 4.30pm)
Royal Statistical Society, Errol Street, London
The effect of indoor air pollutants (volatile organic compounds) on humans
MICHAEL ABRAHAM
(University College, London)
One of the main types of pollutant in indoor air are volatile organic
compounds. The effect of a number of these pollutants on humans has been
determined, but some estimate of effects for a much larger number is
required. The measured effects are threshold levels, that is the minimum
amount of pollutant that can be detected by a person, for eye irritation,
sensory irritation and odor. The effect of these pollutants on humans will
be used as an example of how multiple linear regression analysis is
employed in chemistry to analyze the factors that influence processes, and
to enable predictions of effects to be made. The multiple linear
regression equations obtained for the effect on humans can be compared to
equations for other processes using principle components analysis, in
order to attempt to find 'model' processes that mimic those for effects on
humans. Although the systems studied are chemical in nature, almost no
knowledge of chemistry will be assumed.
Representing compounds for support vector classification
ROBERT BURBIDGE
(Credit Suisse First Boston)
The support vector machine is used to predict whether or not a ligand has
an acceptable binding affinity to various biological receptors. Three
molecular representations are assessed. A novel representation, based on
binding affinities for other receptors, is shown to be superior to the
other representations.
In silico predictions
NIGEL GENSMANTEL
(AstraZeneca)
The development of 'high throughput' drug discovery in terms of chemical
synthesis, screening, physico-chemical characterisation and assessment of
ADME properties has provided the pharmaceutical industry with large
databases of screening and physicochemical data designed to support our
aspirations to develop predictive models of fundamental biological process
at the molecular level. Developments in computer modelling, in silico,
based solely on chemical structure and expert systems for the prediction
of potency, selectivity and pharmacokinetics have the opportunity to
revolutionise the drug discovery process.
ALL ARE WELCOME
For further details of GAS' activities and for travel directions, see
www.rss.org/groups/GAS/
***************************************************************************
Dr Karen Vines (Secretary, GAS)
Department of Statistics
The Open University
Walton Hall
Milton Keynes, MK7 6AA
e-mail: [log in to unmask]
|