# Semi-automated modeling of reaction states in time-resolved serial femtosecond crystallography using molecular dynamics sampled conformations

## TL;DR

This paper introduces a semi-automated method to model molecular structures during chemical reactions using simulations and experimental data.

## Contribution

A new semi-automated approach combining MD simulations and TR-SFX data to model reaction state conformations is introduced.

## Key findings

- MD simulations generate conformational samples used as initial models for TR-SFX data refinement.
- Evaluated models show high-quality geometries and electron density metrics using Edstats and MolProbity.

## Abstract

Time-resolved serial femtosecond crystallography (TR-SFX) is a technique designed to reveal the molecular dynamics underlying chemical reactions, thereby providing insights into the relationship between structure and function. By capturing a series of conformational changes in intermediate states, TR-SFX enables the visualization of dynamic structural transitions. In this study, we have presented a new approach for modeling reaction state conformations using molecular dynamics (MD) simulations. In this approach, MD simulations were first performed to generate a large number of conformational samples, which were then used as initial models for refinement against diffraction data from the triggered states, thereby facilitating the construction of accurate dynamic structure models. The derived models were evaluated using tools such as Edstats and MolProbity to identify models with high-quality geometries and local electron density metrics. This procedure provides a semi-automated approach for building dynamic structural models from TR-SFX data, ensuring their robustness for further exploration and understanding of macromolecular dynamics.