Software

Several of our in-house developments to analyze 'big-data' MD simulations can be obtained from our GitHub page.

$ 

FastPCA

The FastPCA package is an implementation of the principal component analysis of large MD data sets, using either Cartesian atom coordinates, interatom distances or backbone dihedral angles as input coordinates. In particular, it features the dihedral angle PCA on a torus (dPCA+) by Sittel et al., 2017, which performs maximal gap shifting to treat periodic data correctly. It is optimized and parallelized with constant memory consumption for large data sets. Technical information and the source code can be found at the project site.

Clustering

The clustering package is a highly optimized C++11 program suite parallelized with CUDA and OpenMP. It supports for: The latest version and installation instructions can be found on the project site. Additionally, there is a comprehensive step-by-step technical explanation of a full clustering workflow, based on a sample data set. Further information can be found in the documentation.

MoSAIC

MoSAIC (molecular systems automated identification of correlation) is a scalable and unbiased method for detecting collective motion in proteins. It can be used as a feature selection scheme for subsequent modeling or to gain an understanding of the cooperativity in the biomolecular system. The method and some applications are described in Diez et al., 2022. A documentation and the source code can be found at the project site.

xgbAnalysis

The xgbAnalysis package developed by Brandt et al. is a tool to find suitable reaction coordinates of biomolecular systems, e.g. proteins, using the XGBoost algorithm. Given a trajectory from a molecular dynamics simulation and suitable corresponding states, it evaluates, which of the input coordinates describe the system best, resulting in a low dimensional reaction coordinate of directly interpretable original coordinates.

Dissipation-corrected targeted molecular dynamics (dcTMD)

Python scripts used for dissipation-corrected targeted molecular dynamics by Wolf et al., 2018 analysis for usage with "*pullf.xvg" files from Gromacs. More information can be found at the github page.

Data-Driven Langevin Package

We have developed together with R. Hegger a systematic computational approach to describe the conformational dynamics of biomolecules in reduced dimensionality using data-driven Langevin equation modeling.
For details see Hegger et al., 2009 and Schaudinnus et al., 2016.
The software can be downloaded from github.com/moldyn/Data-Driven-Langevin.

Temperature-boosted Langevin equation simulations

Based on free energy and friction profiles from dissipation-corrected targeted MD, we have developed an approach to accelerate the prediction of system kinetics by "temperature boosting“ Markovian Langevin equation simulations. A C++ code for such simulations and Jupiter notebooks for data analysis and error estimates can be found at https://github.com/moldyn/Langevin_T_boost.

Ramacolor

The ramacolor script (written in python3) as proposed in Sittel et al., 2016 can be downloaded directly from here. Ramacolor plots visualize the secondary structures of given microstates. Hence, states can be assigned as dynamically active/inactive at a glance.