Transition state searching for complex biomolecules: Algorithms and machine learning

Transition state is a key concept for chemists to understand and fine-tune the conformational changes of large biomolecules. Due to its short residence time, it is difficult to capture a transition state via experimental techniques. Characterizing transition states for a conformational change therefore is only achievable via physics-driven molecular dynamics simulations. However, unlike chemical reactions which involve only a small number of atoms, conformational changes of biomolecules depend on numerous atoms and therefore the number of their coordinates in our 3D space. The searching for their transition states will inevitably encounter the curse of dimensionality, i.e. the reaction coordinate problem, which invokes the invention of various algorithms for solution. Recent years, new machine learning techniques and the incorporation of some of them into the transition state searching methods emerged. Here, we first review the design principle of representative transition state searching algorithms, including the collective-variable (CV)-dependent gentlest ascent dynamics, finite temperature string, fast tomographic, travelling-salesman based automated path searching, and the CV-independent transition path sampling. Then, we focus on the new version of TPS that incorporates reinforcement learning for efficient sampling, and we also clarify the suitable situation for its application. Finally, we propose a new paradigm for transition state searching, a new dimensionality reduction technique that preserves transition state information and combines gentlest ascent dynamics.