Classical Molecular Dynamics (MD) simulations can provide insights at the nanoscopic scale into protein dynamics. Currently, simulations of large proteins and complexes can be routinely carried out in the ns-µs time regime. Clustering of MD trajectories is often performed to identify selective conformations and to compare simulation and experimental data coming from different sources on closely related systems. However, clustering techniques are usually applied without a careful validation of results and benchmark studies involving the application of different algorithms to MD data often deal with relatively small peptides instead of average or large proteins; finally clustering is often applied as a means to analyze refined data and also as a way to simplify further analysis of trajectories. Herein, we propose a strategy to classify MD data while carefully benchmarking the performance of clustering algorithms and internal validation criteria for such methods. We demonstrate the method on two showcase systems with different features, and compare the classification of trajectories in real and PCA space. We posit that the prototype procedure adopted here could be highly fruitful in clustering large trajectories of multiple systems or that resulting especially from enhanced sampling techniques like replica exchange simulations.

Cherry-picking functionally relevant substates from long md trajectories using a stratified sampling approach

Chandramouli B.;Mancini G.
2016-01-01

Abstract

Classical Molecular Dynamics (MD) simulations can provide insights at the nanoscopic scale into protein dynamics. Currently, simulations of large proteins and complexes can be routinely carried out in the ns-µs time regime. Clustering of MD trajectories is often performed to identify selective conformations and to compare simulation and experimental data coming from different sources on closely related systems. However, clustering techniques are usually applied without a careful validation of results and benchmark studies involving the application of different algorithms to MD data often deal with relatively small peptides instead of average or large proteins; finally clustering is often applied as a means to analyze refined data and also as a way to simplify further analysis of trajectories. Herein, we propose a strategy to classify MD data while carefully benchmarking the performance of clustering algorithms and internal validation criteria for such methods. We demonstrate the method on two showcase systems with different features, and compare the classification of trajectories in real and PCA space. We posit that the prototype procedure adopted here could be highly fruitful in clustering large trajectories of multiple systems or that resulting especially from enhanced sampling techniques like replica exchange simulations.
Settore CHIM/02 - Chimica Fisica
Analysis of trajectories; Cluster validation; Density based clustering; Molecular dynamics; Binding Sites; Cluster Analysis; Principal Component Analysis; Proteins; Algorithms; Molecular Dynamics Simulation; Protein Conformation
File in questo prodotto:
File Dimensione Formato  
Mancini and Chandramouli - 2016 - Cherry-Picking Functionally Relevant Substates fro.pdf

Accesso chiuso

Tipologia: Accepted version (post-print)
Licenza: Non pubblico
Dimensione 2.55 MB
Formato Adobe PDF
2.55 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11384/109048
Citazioni
  • ???jsp.display-item.citation.pmc??? 0
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact