Skip to Main content Skip to Navigation
Theses

Prédiction de la localisation cellulaire des protéines à l'aide de leurs séquences biologiques.

Abstract : Cellular compartments, due to the membrane frontiers they induce, allow the realization of diverse metabolic tasks in the cell. This specialization of the cell's spatial domains directly corresponds to a differentiation in the functional role of the proteins they involve. Thus, when homology searches in the databanks produce no results, the knowledge of the localization site of a protein can help in deducing its function. With the large amount of unannotated orphan genes produced these last years, ab-initio prediction of the subcellular location of proteins has become an important problem.

Thus the major part of the work presented here concerns the prediction of the subcellular localization of a protein, knowing its primary or coding sequence.

We proposed descriptive alternatives to existing methods for predicting subcellular localization by: (1) using new descriptors from nucleotidic sequence and (2) an HMM approach combined with decision trees. The HMM approach is justified biologically in that it permits to modelize biological adressing signals conjointly with global composition. Furthermore, embedding the classification steps within a decision tree slightly improved classification results, whose accuracy is similar with other methods using global composition information.
Document type :
Theses
Complete list of metadata

Cited literature [210 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00011707
Contributor : Hugues Richard <>
Submitted on : Wednesday, March 1, 2006 - 7:57:14 PM
Last modification on : Tuesday, March 17, 2020 - 2:58:01 AM
Long-term archiving on: : Monday, September 17, 2012 - 12:21:02 PM

Identifiers

  • HAL Id : tel-00011707, version 1

Collections

Citation

Hugues Richard. Prédiction de la localisation cellulaire des protéines à l'aide de leurs séquences biologiques.. Mathématiques [math]. Université d'Evry-Val d'Essonne, 2005. Français. ⟨tel-00011707⟩

Share

Metrics

Record views

335

Files downloads

3892