SCIENTIFIC CONTEXT. Biomolecules, particularly proteins and nucleic acids (DNA and RNA), are essential to biological processes. Their study is typically divided into two main aspects: the structural level, which focuses on their (meta-)stable conformations, and dynamics, which evaluates the stability of these conformations and the rates at which they interconvert. AlphaFold, the program developed by Deepmind, made a very significant step for structural predictions of structured proteins, and its two leaders were co-awarded the 2024 Nobel prize in chemistry. But the question of fast and accurate predictions of dynamical properties remains open. This is a particularly important problem to understand the strength of interactions at the molecular level, with potentially far reaching applications in medicine. The goal of the project EMINENCE–molEcular dynaMics In iNtErNal CoordinatEs, funded by France 2030–see below, is to deliver state-of-the-art modeling tools geared towards biomolecular dynamics, within the Structural Bioinformatics Library.

SOFTWARE: the Structural Bioinformatics Library. The Structural Bioinformatics Library(SBL, https: //sbl.inria.fr) is a comprehensive software environment to model biomolecules, with a unique and versatile design. It jointly provides (i) low level algorithm classes, (ii) biophysical models, and (iii) applications solving specific problems in structural bioinformatics. Applications are ascribed to two principal categories respectively targeting static and dynamic models, see https://sbl.inria.fr/applications/. To foster genericity, re-usability, efficiency and robustness, the SBL is mainly developed in generic (template based) C++, and consists of ∼ 100 C++ packages, 2/3 in the Core, and 1/3 for Applications.

FRANCE 2030. France 2030 is an investment plan aiming at supporting innovative technologies and the ecological transition https://www.economie.gouv.fr/france-2030. The project EMINENCE is supported by the Inria Quadrant Program and receives national funding through The French National Research Agency as part of the France 2030 plan under the reference « ANR24-RRII-0002 », operated by the Inria Quadrant Program. In the framework of France 2030 and the Digital program agency Digital Program Agency - Algorithms, software and usage operated by Inria, the Inria Quadrant Program selected by the General Secretariat for Investment (SGPI) and the Directorate-General for Research and Innovation (DGRI) aims to support scientific risk-taking and to address current and future challenges in research and innovation for and through digital sciences and technology. See https://piq.inria.fr/.

ENGINEERING POSITION AND TASKS: ALGORITHM ENGINEERING FOR BIOMOLECULAR DYNAMICS.

In modeling biomolecules, internal coordinates in general and torsion angles in particular are much more effective than Cartesian coordinates to generate large amplitude conformational changes. However, using torsion angles is more challenging mathematically, in several respects: necessity to solve inverse problems of the loop closure type, non local effects, sampling biases. We have recently developed several algorithms to explore energy landscapes, and sample loops as well as multiloop systems.

* Main goals/specific tasks
** Optimizing loop and multiloop samplers.
** Optimizing energy landscape explorers.
** Optimizing binding affinity estimators.