Skip to Main content Skip to Navigation
Theses

Prédiction de structures secondaires d’ARN et de complexes d’ARN avec pseudonoeuds - Approches basées sur la programmation mathématique multi-objectif

Abstract : In this thesis, we propose new algorithms and tools to predict RNA and RNA complex secondary structures, including particular RNA motifs, difficult to predict, like pseudoknots. RNA structure prediction stays a difficult task, and the numerous existing tools don't always give good predictions. In order to predict structures that are as close as possible to the real ones, we propose to develop algorithms that:i) predict the k-best structures;ii) combine several models of prediction to take advantage of each;iii) are able to take into account user constraints and structural data like SHAPE.We developed three tools: BiokoP for predicting RNA secondary structures and RCPred and C-RCPred for predicting RNA complex secondary structures. The tool BiokoP proposes several optimal and sub-optimal structures thanks to the combination of two prediction models, the energy model MFE and the probabilistic model MEA. This combination is done with multi-objective mathematical programming, where each model is associated to an objective function.To this end, we developed a generic algorithm returning the k-best Pareto curves of a bi-objective integer linear program. The tool RCPred, based on the MFE model, proposes several sub-optimal structures. It takes advantage of the numerous existing tools for RNA secondary structure prediction and for RNA-RNA interaction prediction, by taking as input predicted secondary structures and RNA-RNA interactions. The goal of RCPred is to find the best combination among these inputs. The tool C-RCPred is a new version of RCPred, taking into account user constraints and structural data (SHAPE, PARS, DMS). C-RCPred is based on a multi-objective algorithm, where the different objectives are the MFE model, the fulfillment of the user constraints and the concordance with the structural data.
Document type :
Theses
Complete list of metadata

Cited literature [261 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/tel-02813758
Contributor : Frédéric Davesne <>
Submitted on : Saturday, June 6, 2020 - 11:36:05 AM
Last modification on : Friday, July 17, 2020 - 10:44:23 AM

File

2019SACLE031.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02813758, version 1

Collections

Citation

Audrey Legendre. Prédiction de structures secondaires d’ARN et de complexes d’ARN avec pseudonoeuds - Approches basées sur la programmation mathématique multi-objectif. Bio-informatique [q-bio.QM]. Université Paris-Saclay, 2019. Français. ⟨NNT : 2019SACLE031⟩. ⟨tel-02813758⟩

Share

Metrics

Record views

78

Files downloads

164