IIIT Hyderabad Publications |
|||||||||
|
Augmenting MD simulation studies of functional RNAs using network and data mining based approaches – a case study with add Adenine riboswitchAuthor: Rama Dedeepya Payila Date: 2020-06-12 Report no: IIIT/TH/2020/48 Advisor:Abhijit Mitra AbstractKnowledge of structure and dynamics of biomolecules is important to understand their functional mechanism. While X-Ray crystal structures provide information about the structural details, atomic level details related to inherent flexibilities and functional dynamics of the biomolecule cannot be obtained from them. Therefore molecular dynamics (MD) simulations are widely used to gain a molecular level understanding of the functional mechanism of biomolecules. Trajectory generated from MD simulation often consists of large number of snapshots or conformations. Each conformation in the trajectory contains three dimensional cartesian coordinates of all the atoms in the system making MD data huge that can be difficult to analyze. In the present study, a network based approach (Base Correlation Network) to have a simplified representation of MD data that can capture the necessary information and data mining approaches (Classification and Frequent Itemset Mining) to extract the relevant information from MD data were employed. These methods can be applied to riboswitches to understand the differential dynamics of its ligand free and ligand bound states in order to gain insights into their functioning. The methods have been tested on the ligand free and ligand bound aptamer domain of add Adenine riboswitch. Base Correlation Networks (BCN) were constructed based on cross correlation data obtained from molecular dynamics simulations, for ligand bound and ligand free states of add Adenine riboswitch. These networks have been able to identify the crucial residues in each state. Comparison of communities (correlated regions) between the ligand free (OPEN) and ligand bound (CLOSED) state BCNs revealed how the ligand binding affected the correlated motions within the aptamer. Also, a classification algorithm was used to identify the structural features that discriminate between the ligand free and ligand bound states of aptamer domain in an automated fashion. This method revealed the shortening of P1 helix and tightening of L2-L3 region on ligand binding. In addition, this method also identified base interactions and base distances that show significant differences between the ligand free and ligand bound states. Further, Frequent Itemset Mining (FIM) was employed to understand the role of the ligand binding pocket and the L2-L3 regions, in terms of molecular level interactions, both in the presence and absence of ligand, without losing the time dependent correlation between the base interactions. This method also provided insights into ligand binding mechanism. While all the three methods have been successfully applied to the add Adenine riboswitch aptamer, to understand the differential dynamics between their ligand bound and ligand free states; in particular, network and frequent itemset mining methods can be applied to other functional RNAs in order to understand their functioning. Full thesis: pdf Centre for Computational Natural Sciences and Bioinformatics |
||||||||
Copyright © 2009 - IIIT Hyderabad. All Rights Reserved. |