| Doctorant : |
Iyanar VETRIVEL
|
| Directeur de thèse : |
Bernard OFFMANN ,
Professeur Université |
| co-directeur de thèse : | Frédéric Cadet, Professeur, Université de La Réunion |
| Financement : |
Région Réunion |
| Date de la soutenance : |
lundi 30 octobre 2017, 10h00 |
| Modalité : |
- Lieu : Amphithéâtre Pasteur, bâtiment 2, campus Lombarderie
|
| Jury : |
- Président de jury : Catherine Etchebest, Professeure, Université Paris Diderot
- Rapporteur : Anne-Claude Camproux, Professeure, Université Paris Diderot
- Rapporteur : Juliette Martin, Chargée de Recherche CNRS, Université de Lyon
- Examinateur : Srinivasan Narayanaswamy, Professeur, Indian Institute of Science, Inde
- Directeur de thèse :
Bernard OFFMANN ,
Professeur Université
- co-directeur de thèse : Frédéric Cadet, Professeur, Université de La Réunion
|
Structural motifs found in protein structures. They help in approximating protein structure as a 1D string with minimal loss of structural information. Here I have employed a widely used structural alphabet called Protein Blocks (PB) for various applications like predicting, comparing and analyzing protein structures. PBs were used to study the structural variations in proteins with identical primary structure and also during the course of molecular dynamics (MD) simulations. The results from these analyses were summarized in the form of substitution matrices and showed striking similarities to previously established matrices for homologous proteins and NMR ensembles. I improved kPRED, a knowledge-based prediction of protein backbone in terms of PBs, by taking into consideration the neighboring local structures. The new version of the algorithm also privileges structural information from homologous proteins when available reaching an average accuracy of 66.3% on a benchmark dataset. The scope of the PENTAdb database has been expanded to cover the entire protein structure space in an automated manner. I show that the effect of this 950% increase in the contents of PENTAdb improves our understanding of the sequence-structure relationship at the pentapeptide level. I also used PB-ALIGN, a fast and efficient protein structure comparison tool, to compare all protein structures in PDB in an all-vs-all manner and to investigate PB-based structural similarities. This generated a huge collection of alignment data and I discuss its use for functional annotation and identification of possible evolutionary relationships.
Publications
2019
Vetrivel, Iyanar; Hoffmann, Lionel; Guegan, Sean; Offmann, Bernard; Laurent, Adele D
PBmapclust: Mapping and Clustering the Protein Conformational Space Using a Structural Alphabet Inproceedings
Dans: Byska, Jan; Krone, Michael; Sommer, Björn (Ed.): Workshop on Molecular Graphics and Visual Analysis of Molecular Data, The Eurographics Association, 2019, ISBN: 978-3-03868-085-7.
@inproceedings{lva.20191097b,
title = {PBmapclust: Mapping and Clustering the Protein Conformational Space Using a Structural Alphabet},
author = {Iyanar Vetrivel and Lionel Hoffmann and Sean Guegan and Bernard Offmann and Adele D Laurent},
editor = {Jan Byska and Michael Krone and Björn Sommer},
doi = {10.2312/molva.20191097},
isbn = {978-3-03868-085-7},
year = {2019},
date = {2019-01-01},
booktitle = {Workshop on Molecular Graphics and Visual Analysis of Molecular Data},
publisher = {The Eurographics Association},
abstract = {Analyzing the data from molecular dynamics simulation of biological macromolecules like proteins is challenging. We propose a simple tool called PBmapclust that is based on a well established structural alphabet called Protein blocks (PB). PBs help in tracing the trajectory of the protein backbone by categorizing it into 16 distinct structural states. PBmapclust provides a time vs. amino acid residue plot that is color coded to match each of the PBs. Color changes correspond to structural changes, giving a visual overview of the simulation. Further, PBmapclust enables the user to "map" the conformational space sampled by the protein during the MD simulation by clustering the conformations. The ability to generate sub-maps for specific residues and specific time intervals allows the user to focus on residues of interest like for active sites or disordered regions. We have included an illustrative case study to demonstrate the utility of the tool. It describes the effect of the disordered domain of a HSP90 co-chaperone on the conformation of its active site residues. The scripts required to perform PBmapclust are made freely available under the GNU general public license.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Analyzing the data from molecular dynamics simulation of biological macromolecules like proteins is challenging. We propose a simple tool called PBmapclust that is based on a well established structural alphabet called Protein blocks (PB). PBs help in tracing the trajectory of the protein backbone by categorizing it into 16 distinct structural states. PBmapclust provides a time vs. amino acid residue plot that is color coded to match each of the PBs. Color changes correspond to structural changes, giving a visual overview of the simulation. Further, PBmapclust enables the user to "map" the conformational space sampled by the protein during the MD simulation by clustering the conformations. The ability to generate sub-maps for specific residues and specific time intervals allows the user to focus on residues of interest like for active sites or disordered regions. We have included an illustrative case study to demonstrate the utility of the tool. It describes the effect of the disordered domain of a HSP90 co-chaperone on the conformation of its active site residues. The scripts required to perform PBmapclust are made freely available under the GNU general public license.
Vetrivel, Iyanar; de Brevern, Alexandre G; Cadet, Frédéric; Srinivasan, Narayanaswamy; Offmann, Bernard
Structural variations within proteins can be as large as variations observed across their homologues Article de journal
Dans: Biochimie, vol. 167, p. 162–170, 2019, ISSN: 61831638.
@article{Vetrivel2019,
title = {Structural variations within proteins can be as large as variations observed across their homologues},
author = {Iyanar Vetrivel and Alexandre G de Brevern and Frédéric Cadet and Narayanaswamy Srinivasan and Bernard Offmann},
doi = {10.1016/j.biochi.2019.09.013},
issn = {61831638},
year = {2019},
date = {2019-01-01},
journal = {Biochimie},
volume = {167},
pages = {162--170},
abstract = {Understanding the structural plasticity of proteins is key to understanding the intricacies of their functions and mechanistic basis. In the current study, we analyzed the available multiple crystal structures of the same protein for the structural differences. For this purpose we used an abstraction of protein structures referred as Protein Blocks (PBs) that was previously established. We also characterized the nature of the structural variations for a few proteins using molecular dynamics simulations. In both the cases, the structural variations were summarized in the form of substitution matrices of PBs. We show that certain conformational states are preferably replaced by other specific conformational states. Interestingly, these structural variations are highly similar to those previously observed across structures of homologous proteins (r2 = 0.923) or across the ensemble of conformations from NMR data (r2 = 0.919). Thus our study quantitatively shows that overall trends of structural changes in a given protein are nearly identical to the trends of structural differences that occur in the topologically equivalent positions in homologous proteins. Specific case studies are used to illustrate the nature of these structural variations.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Understanding the structural plasticity of proteins is key to understanding the intricacies of their functions and mechanistic basis. In the current study, we analyzed the available multiple crystal structures of the same protein for the structural differences. For this purpose we used an abstraction of protein structures referred as Protein Blocks (PBs) that was previously established. We also characterized the nature of the structural variations for a few proteins using molecular dynamics simulations. In both the cases, the structural variations were summarized in the form of substitution matrices of PBs. We show that certain conformational states are preferably replaced by other specific conformational states. Interestingly, these structural variations are highly similar to those previously observed across structures of homologous proteins (r2 = 0.923) or across the ensemble of conformations from NMR data (r2 = 0.919). Thus our study quantitatively shows that overall trends of structural changes in a given protein are nearly identical to the trends of structural differences that occur in the topologically equivalent positions in homologous proteins. Specific case studies are used to illustrate the nature of these structural variations.
2017
Vetrivel, Iyanar; Mahajan, Swapnil; Tyagi, Manoj; Hoffmann, Lionel; Sanejouand, Yves-Henri; Srinivasan, Narayanaswamy; Brevern, Alexandre G De; Cadet, Frédéric; Offmann, Bernard
Knowledge-based prediction of protein backbone conformation using a structural alphabet Article de journal
Dans: PLoS ONE, vol. 12, no. 11, 2017, ISSN: 19326203.
@article{Vetrivel2017,
title = {Knowledge-based prediction of protein backbone conformation using a structural alphabet},
author = {Iyanar Vetrivel and Swapnil Mahajan and Manoj Tyagi and Lionel Hoffmann and Yves-Henri Sanejouand and Narayanaswamy Srinivasan and Alexandre G {De Brevern} and Frédéric Cadet and Bernard Offmann},
doi = {10.1371/journal.pone.0186215},
issn = {19326203},
year = {2017},
date = {2017-11-01},
journal = {PLoS ONE},
volume = {12},
number = {11},
publisher = {Public Library of Science},
abstract = {Libraries of structural prototypes that abstract protein local structures are known as structural alphabets and have proven to be very useful in various aspects of protein structure analyses and predictions. One such library, Protein Blocks, is composed of 16 standard 5-residues long structural prototypes. This form of analyzing proteins involves drafting its structure as a string of Protein Blocks. Predicting the local structure of a protein in terms of protein blocks is the general objective of this work. A new approach, PB-kPRED is proposed towards this aim. It involves (i) organizing the structural knowledge in the form of a database of pentapeptide fragments extracted from all protein structures in the PDB and (ii) applying a knowledge-based algorithm that does not rely on any secondary structure predictions and/ or sequence alignment profiles, to scan this database and predict most probable backbone conformations for the protein local structures. Though PB-kPRED uses the structural information from homologues in preference, if available. The predictions were evaluated rigorously on 15,544 query proteins representing a non-redundant subset of the PDB filtered at 30% sequence identity cut-off. We have shown that the kPRED method was able to achieve mean accuracies ranging from 40.8% to 66.3% depending on the availability of homologues. The impact of the different strategies for scanning the database on the prediction was evaluated and is discussed. Our results highlight the usefulness of the method in the context of proteins without any known structural homologues. A scoring function that gives a good estimate of the accuracy of prediction was further developed. This score estimates very well the accuracy of the algorithm (R2 of 0.82). An online version of the tool is provided freely for non-commercial usage at http://www.bo-protscience.fr/kpred/.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Libraries of structural prototypes that abstract protein local structures are known as structural alphabets and have proven to be very useful in various aspects of protein structure analyses and predictions. One such library, Protein Blocks, is composed of 16 standard 5-residues long structural prototypes. This form of analyzing proteins involves drafting its structure as a string of Protein Blocks. Predicting the local structure of a protein in terms of protein blocks is the general objective of this work. A new approach, PB-kPRED is proposed towards this aim. It involves (i) organizing the structural knowledge in the form of a database of pentapeptide fragments extracted from all protein structures in the PDB and (ii) applying a knowledge-based algorithm that does not rely on any secondary structure predictions and/ or sequence alignment profiles, to scan this database and predict most probable backbone conformations for the protein local structures. Though PB-kPRED uses the structural information from homologues in preference, if available. The predictions were evaluated rigorously on 15,544 query proteins representing a non-redundant subset of the PDB filtered at 30% sequence identity cut-off. We have shown that the kPRED method was able to achieve mean accuracies ranging from 40.8% to 66.3% depending on the availability of homologues. The impact of the different strategies for scanning the database on the prediction was evaluated and is discussed. Our results highlight the usefulness of the method in the context of proteins without any known structural homologues. A scoring function that gives a good estimate of the accuracy of prediction was further developed. This score estimates very well the accuracy of the algorithm (R2 of 0.82). An online version of the tool is provided freely for non-commercial usage at http://www.bo-protscience.fr/kpred/.
Lien
