Lasso Folding

This repository contains all the source code for analyzing the lasso folding molecular dynamics simulation data of 20 different lasso peptides.

Features

• Featurization of MD trajectory data
• Construction and validation of Markov State Models (MSM)
• Thermodynamic and kinetic analysis
• Identification and clustering of kinetic folding pathways

Repository Structure

FeatureCalculation_and_MSM

Contains scripts to:

• Featurize MD trajectory data using all pair-wise residue-residue distances
  • feature_calculation.py: Calculate all pair-wise residue-residue distances of all trajectory data.
• Construct and validate Markov State Models for each lasso peptide
  • optimize_MSM_hyperparameter.py: Perform a grid search to optimize the hyperparameter of MSM based on VAMP2 score and construct MSM using each hyperparameter combination.
  • MSM_reweighted_population_analysis.py: Evaluate the stationary distribution from the MSM and its deviation from the original distribution.
  • MSM_population_of_cluster.py: Evaluate the stationary distribution of each cluster.
  • bootstrapping_MSM.py: Estimate relative errors through bootstrapping. 80% of trajectories are randomly selected for each bootstrap sample, and MSM is constructed for each sample.

Thermodynamics_Kinetics

Contains scripts to perform thermodynamic and kinetic analysis of the 20 lasso peptides.

• Thermodynamics analysis
  • NativeContacts_calculation.py: Calculate the fraction of native contacts of each microstate (cluster).
  • thermodynamics_analysis.py: Calculate the percentage of lasso-like topology formation (Native Contacts > 0.8).
• Kinetic analysis
  • kinetics_analysis.py: Estimate the transition timescale from unfolded states to folded states by MFPT (mean first passage time).
• Plot MSM-weighted free energy landscape
  • Free_energy_landscape_plot.py: Project MD trajectory data on tIC 1 and tIC2, and project the average native contact value of each microstate on the landscape.

Kinetic_pathways_analysis

Contains scripts to identify and cluster kinetic folding pathways.

• kinetic_pathway_from_TPT.py: Obtain all kinetic pathways from unfolded to folded state based on TPT (Transition Path Theory).
• project_pathway_on_tICA_space.py: Project each pathway on low dimensional subspace (tICA space).
• train_VAE_cluster_pathways.py: Train VAE model to learn the latent space of each pathway.
• optimize_cluster_number_by_silhouette_analysis.py: Optimize the cluster number used for performing k-means clustering on VAE latent space of all pathways.
• plot_pathways_on_latent_space.py: Visualize the clustering results of all pathways on VAE latent space and plot train-validation loss of VAE model.

Dependency

To set up the environment for this project, use the provided environment.yml file. This file contains all necessary dependencies.

Authors

• Xuenan Mi - [email protected]

Contributing

Contributions are welcome! Please create an issue to discuss the changes you wish to make.