The Development of More Accurate QSAR Techniques
QSAR is a very effective starting step in the development of compounds for vast numbers of industries. Its scale and importance, especially in the medicinal field means it is a dynamic area to research. The size of QSAR also presents problems; there are many different methods in use for each of the...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2012 |
| País: | Argentina |
| Institución: | Consejo Nacional de Investigaciones Científicas y Técnicas |
| Repositorio: | CONICET Digital (CONICET) |
| Idioma: | inglés |
| OAI Identifier: | oai:ri.conicet.gov.ar:11336/103247 |
| Acceso en línea: | http://hdl.handle.net/11336/103247 |
| Access Level: | acceso abierto |
| Palabra clave: | Qsar Theory Replacement Method K-Means Clustering Molecular Descriptors https://purl.org/becyt/ford/1.4 https://purl.org/becyt/ford/1 |
| Sumario: | QSAR is a very effective starting step in the development of compounds for vast numbers of industries. Its scale and importance, especially in the medicinal field means it is a dynamic area to research. The size of QSAR also presents problems; there are many different methods in use for each of the steps in a study, from the descriptors in use, to the type of linear regression to apply to the descriptors. The idea was to put forward models that improved upon the existing methods to such a degree that it could become a universal method for QSAR modelling. This project successfully investigated in detail an improvement to the existing methods to choose the correct number of descriptors to include in the model by using Rloo analysis; this resulted in a simpler model compared to previous methods. K – Means clustering was also investigated as part of a novel, variable independent method. This methodology only uses one descriptor as opposed to general QSAR studies which use several. The results for 12 out of the 14 sets were at least as accurate as the results obtained by existing linear methods. An example using PERM; the Stest obtained using the novel method was 0.46 compared to the Stest of 0.53 obtained by using current linear methods. The simplicity associated with the K - Means clustering method and the fact it shows improved predictive potential could lead to an overhaul of all current, more complicated methods in favour of the simpler K- Means based method. |
|---|