Mots-Clés

Protein-Nucleic Acid Conformational dynamics Graph analysis

Description

Context

The main goal of this Postdoc is to elucidate the conformational dynamics of protein-nucleic acid complexes and to develop a method for understanding their communications. To achieve this objective, the candidate will develop a method to predict communication networks. They will investigate several facets of the algorithms and ultimately provide this application to the broad scientific community. The proposed deep learning model will be set up to address the characterization of dynamic properties of protein-nucleic acid complexes at the 3D level. This approach is a prerequisite for the design of new therapeutics that target protein-nucleic acid complexes.

The candidate will be hosted in the CAPSID team within LORIA and Inria center at the Universite de Lorraine. The candidate will be supervised by Yasaman Karami (Inria researcher) with expertise in proteins conformational dynamics and drug design [1,2]. CAPSID team (https://capsid.loria.fr/) is directed by Dr. Marie-Dominique Devignes and provides a multidisciplinary and international en- vironment for students. The team benefits from experts in RNA-protein interactions and structural bioinformatics, as well as in computer science. The group is equipped with a computational plat- form, MBI-DS4H (https://mbi-ds4h.loria.fr/) composed of 8 nodes and 12 GPUs, and provides technical support to the users.

Background

Biomolecules such as proteins and nucleic acids are at the heart of virtually all funda- mental cellular processes. They adopt complex dynamic behavior and their functions are directly linked to the arrangement of atoms in 3D (structure) and dynamics. These systems undergo conformational changes in response to different environmental conditions, such as mutations, changes in temperature or electrostatic potential, and binding to other molecules. Such changes may alter the structural plas- ticity of a biomolecule, induce malfunctioning, thereby provoking diseases. Therefore, characterizing the structure, dynamics and conformational changes of biomolecules can help understand the molec- ular mechanisms of underlying diseases, and hopefully prevent them by designing drugs. Molecular modeling is the field of characterizing structure of biomolecules. While major improvements have been introduced with the breakthrough of artificial intelligence (AlphaFold2 [3]), predicting conformational dynamics of biomolecules is still challenging. At the same time, molecular dynamics (MD) simula- tions can provide a powerful, versatile, and accurate computational microscope to study the dynamic behavior of biological systems. These simulations generate conformational ensembles to capture the motion of systems. Considering the dynamic nature of protein-nucleic acid complexes and the role of dynamics in the formation of such interactions, having a single frozen structural state (such as the protein models currently obtained by deep learning) is not sufficient to fully understand the molecular mechanisms governing the behavior of such complexes and to modulate their functions.

Main task

In this project the goal is to design and implement a method for characterizing the conformational dynamics of protein-nucleic acid complexes using the set of trajectories generated from MD simulations in a systematic way. This is of high importance specifically due to the complexity of these systems and large size of produced data from MD simulations. On the one hand, MD simulations have been largely employed to analyse protein-protein complexes, however their application in protein- nucleic acid complexes has yet to be explored. The performance of MD simulations directly depends on the force-field parameters. The recent improvements of such parameters have made them suitable for the analysis of nucleic acids and for characterizing conformational changes in DNA/RNA molecules. On the other hand, computational approaches have been developed to investigate the conformational dynamics in proteins [4,5,6]. In the proposed project, the main goal is to adapt such methods to analyse protein-nucleic acid complexes.

Main activities

• Literature review of the relevant studies

• Developing a method to extract communication networks in protein-nucleic acid complexes

• Implementing the method and preparing a software using Python

• Validating the method and analysing the results

• Writing scientific articles and presenting the work in international conferences

Skills

• PhD in Computer Science, Bioinformatics, Chemoinformatics or a related program

• Proficiency in programming languages (Python, R, C++) and good coding practices

• Skills in algorithm design and graph theory

• Ability to work independently and also to work in a team

• Excellent oral and written English skills

References

[1] Temmam S, Vongphayloth K, Baquero E, Munier S, Bonomi M, Regnault B, Douangboubpha B, Karami Y, et al. Bat coronaviruses related to SARS-CoV-2 and infectious for human cells. Nature. 2022; 604.7905: 330-336.

[2] Karami Y, Lopez-Castilla A, Ori A, Thomassin JL, Bardiaux B, Malliavin T, Izadi-Pruneyre N, Francetic O, Nilges M. Computational and biochemical analysis of type IV pilus dynamics and stability. Structure. 2021; 29.12: 1397-1409.

[3] Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Zidek A, Potapenko A, Bridgland A. Highly accurate protein structure prediction with AlphaFold. Nature. 2021 Aug 26;596(7873):583-9.

[4] Gheeraert A, Vuillon L, Chaloin L, Moncorge O, Very T, Perez S, Leroux V, Chauvot de Beauch- ene I, Mias-Lucquin D, Devignes MD, Rivalta I. Singular Interface Dynamics of the SARS-CoV-2 Delta Variant Explained with Contact Perturbation Analysis. Journal of Chemical Information and Modeling. 2022 Jun 6;62(12):3107-22.

[5] Karami Y, Laine E, Carbone A. Dissecting protein architecture with communication blocks and communicating segment pairs. BMC bioinformatics. 2016; 17.2: 133-148.
[6] Karami Y, Bitard-Feildel T, Laine E, Carbone A.“Infostery”analysis of short molecular dynamics simulations identifies highly sensitive residues and predicts deleterious mutations. Scientific reports 2018; 8(1): 1-18.