Selecting Features for Markov Modeling: A Case Study on HP35

This repository provides all the scripts and intermediate steps to reproduce the analysis of Nagel et al. 2023. If the provided scripts/files are used, please cite:

Selecting Features for Markov Modeling: A Case Study on HP35,
D. Nagel, S. Sartore, and G. Stock,
J. Chem. Theory Comput. 2023, ASAP
doi: 10.1021/acs.jctc.3c00240

Getting Started

DEPENDENCY: git-lfs is needed for large file support

To download all the included submodules, please clone this repository with

git clone --recurse-submodules [email protected]:moldyn/HP35.git
cd HP35

To keep the prerequisites as low as possible, we use the slower CPU implementation of density-based clustering here.

Create States

Features: Backbone Dihedral Angles and Minimal Contact Distances

In the directory HP35-DESRES you can find

1. hp35.dihs: backbone dihedral angles [degrees]
2. hp35.dihs.shifted: maximum-gap shifted backbone dihedral angles [rad]
3. hp35.crystaldists: the atom distances of all contacts occurring in the crystal structure 2f4k [nm]
4. hp35.mindists: all minimal distances occurring more frequently than 30% of the time [nm]
5. hp35.mindists2: improved contact distances definition with all atom pairwise distances occurring more frequently than 30% of the time [nm]

For more details, take a look at the repository HP35-DESRES. For legal reasons, the data cannot be integrated directly. Please note the attached license when using it.

Principal Components

In the directory PCA you can find the resulting principal component projections

1. hp35.dihs.res3-33.shifted.gaussian10f.proj.1-4
2. hp35.mindists2.gaussian10f.proj.1-5 In the same directory you can also find a script to reproduce them. For more information please take a look at the README.

Microstate Trajectories

In the directory CLUSTERING you can find the resulting microstate trajectories.

1. hp35.dihs.res3-33.shifted.gaussian10f_microstates_pcs4_p153
2. hp35.mindist2.gaussian10f_microstates_pcs5_p153

In the same directory you can also find a script to reproduce them. For more information please take a look at the README.

Macrostate Trajectories

In the directory MPP you can find the resulting macrostate trajectories.

1. hp35.dihs.res3-33.shifted.gaussian10f_microstates_pcs4_p153.mpp50_transitions.dat.renamed_by_q.pop0.001_qmin0.50.macrotraj
2. hp35.mindists2.gaussian10f_microstates_pcs5_p153.mpp50_transitions.dat.renamed_by_q.pop0.005_qmin0.50.macrotraj_lumped13

In the same directory you can also find a script to reproduce them. For more information please take a look at the README.

Markov State Analysis

In the directory MSM you can find a description of how to reproduce the Markov state model analysis of Nagel et al. 2023. All the analysis is based on the Python package msmhelper.

CK-test	Kinetic Network	Contact Rep.