IGLOO: An Iterative Global Exploration and Local Optimization Algorithm to Find Diverse Low-Energy Conformations of Flexible Molecules

Abstract

The exploration of the energy landscape of a chemical system is essential for understanding and predicting its observable properties. In most cases, this is a challenging task due to the high complexity of such landscapes, which often consist of multiple, possibly hierarchical basins that are difficult to locate and thoroughly explore. In this study, we introduce a novel method, called IGLOO (Iterative Global Exploration and Local Optimization), which aims to achieve a more efficient global exploration of the conformational space compared to existing techniques. The method utilizes a tree-based exploration inspired by the Rapidly exploring Random Tree (RRT) algorithm originating from robotics. IGLOO dynamically adjusts its exploration strategy to both homogeneously scan the landscape and focus on promising regions, avoiding redundant exploration. We evaluated IGLOO using models of two polypeptides and compared its performance to the traditional basin-hopping method and a hybrid method that also incorporates the RRT algorithm. We find that IGLOO outperforms both alternative methods in terms of efficiently and comprehensively exploring the molecular conformational space. This approach can be easily generalized to other chemical systems such as molecules on surfaces or crystalline systems.

Exploring Molecular Energy Landscapes by Coupling the DFTB Potential with a Tree-Based Stochastic Algorithm: Investigation of the Conformational Diversity of Phthalates

“…A more detailed description of these stages is provided below, together with the implementation details. More in-depth explanations about IGLOO can also be found in a recent work, which demonstrates the good performance of the algorithm compared to related methods for finding low-energy conformations of atomic or molecular systems. Note also that IGLOO was successfully applied to predict the structure of disaccharide molecules on metal surfaces. , …”

“…After the initialization, the algorithm iterates three successive stages (see ref for deeper explanations on the method): Exploration : At each iteration, IGLOO explores the conformational space using a stochastic process starting from a set of states. For the first iteration, these states are randomly sampled using a strategy inspired from the Poisson disk sampling process to ensure good dispersion.…”

“…One of these algorithms, known as rapidly exploring random trees (RRT), was subsequently extended to the exploration of energy landscapes aiming to find transition paths. − More recently, the RRT and BH algorithms have been combined to find energy minima on a PES . The strategy applied in this work, called iterative global exploration and local optimization (IGLOO), iterates RRT-based exploration and local minimization. The IGLOO algorithm will be detailed in Materials and Methods.…”