Objectif
A new methodology for the analysis of data obtained from n-dimensional systems will be developed from different perspectives:
(i) optimisation,
(ii) data analysis and
(iii) automation. Approaches (i) and (ii) are strongly related: the former will be focuse d on finding the best experimental conditions for the latter.
Part (i). We will develop new resolution functions for optimisation purposes in n-dimensional separations. The basis of this new resolution function will be concept of Net Analyte Signal (NAS).
Part (ii). A new philosophy for data analysis is proposed. Instead of decomposing the matrix of data into sources of mathematical variance, a new concept, called "chemical variance " will be used.
This new approach will focus on the computation of the dev iations of the data from chemical retention models. This will avoid the constraint of bilinearity, which is an essential requirement when common multivariate techniques (e.g. PLS, PARAFAC) are applied. This approach will be applied to the analysis of very complex mixtures, for which the property of bilinearity does not always hold (e.g. distributions of polymers, biosystems).
Part (iii). The last part of the project will be devoted to the automation of the process. The aim is to build a software package ori ented to be used by non-experienced users, broadening the applicability of the ideas developed in the proposal.
Champ scientifique
Appel à propositions
FP6-2004-MOBILITY-5
Voir d’autres projets de cet appel