Limits...
Automatic identification of mobile and rigid substructures in molecular dynamics simulations and fractional structural fluctuation analysis.

Martínez L - PLoS ONE (2015)

Bottom Line: These substructures are automatically identified by the method.The algorithm consists of the iterative superposition of the fraction of structure displaying the smallest displacements.Examples are given to illustrate the interpretative advantages of this strategy.

View Article: PubMed Central - PubMed

Affiliation: Institute of Chemistry, University of Campinas-UNICAMP, Campinas, SP, Brazil.

ABSTRACT
The analysis of structural mobility in molecular dynamics plays a key role in data interpretation, particularly in the simulation of biomolecules. The most common mobility measures computed from simulations are the Root Mean Square Deviation (RMSD) and Root Mean Square Fluctuations (RMSF) of the structures. These are computed after the alignment of atomic coordinates in each trajectory step to a reference structure. This rigid-body alignment is not robust, in the sense that if a small portion of the structure is highly mobile, the RMSD and RMSF increase for all atoms, resulting possibly in poor quantification of the structural fluctuations and, often, to overlooking important fluctuations associated to biological function. The motivation of this work is to provide a robust measure of structural mobility that is practical, and easy to interpret. We propose a Low-Order-Value-Optimization (LOVO) strategy for the robust alignment of the least mobile substructures in a simulation. These substructures are automatically identified by the method. The algorithm consists of the iterative superposition of the fraction of structure displaying the smallest displacements. Therefore, the least mobile substructures are identified, providing a clearer picture of the overall structural fluctuations. Examples are given to illustrate the interpretative advantages of this strategy. The software for performing the alignments was named MDLovoFit and it is available as free-software at: http://leandro.iqm.unicamp.br/mdlovofit.

No MeSH data available.


Related in: MedlinePlus

Analysis of the mobility of two simulations of a Burkholderia cepacia lipase [14] in mixtures of water and sorbitol.Different RMSD profiles are observed: (A) The standard RMSD of Simulation 1 (black) is much lower than the RMSD of Simulation 2 (red). (B) RMSD as a function of the fraction of the atoms considered in the alignment. These plots indicate that in Simulation 2 (red), there is a subset of about 25 to 30% of the atoms which are responsible for the greater overall RMSD observed in panel (A). (C) In both simulations, 70% of the atoms can be superposed to less than 1Å (RMSDL—dotted lines, black for Simulation 1, red for Simulation 2). The remaining 30% of the atoms behave differently in each simulation (RMSDH—solid lines, same colors). (D) and (E) Superposition of the frames and coloring of the 70% least mobile atoms (blue) and 30% most mobile atoms (red) provides the structural basis for the differential RMSDs. (F) Structures of Simulation 2 colored according to the RMSF of each residue relative to the initial structure after alignment. All these plots and figures can be obtained from the output of MDLovoFit.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4376797&req=5

pone.0119264.g002: Analysis of the mobility of two simulations of a Burkholderia cepacia lipase [14] in mixtures of water and sorbitol.Different RMSD profiles are observed: (A) The standard RMSD of Simulation 1 (black) is much lower than the RMSD of Simulation 2 (red). (B) RMSD as a function of the fraction of the atoms considered in the alignment. These plots indicate that in Simulation 2 (red), there is a subset of about 25 to 30% of the atoms which are responsible for the greater overall RMSD observed in panel (A). (C) In both simulations, 70% of the atoms can be superposed to less than 1Å (RMSDL—dotted lines, black for Simulation 1, red for Simulation 2). The remaining 30% of the atoms behave differently in each simulation (RMSDH—solid lines, same colors). (D) and (E) Superposition of the frames and coloring of the 70% least mobile atoms (blue) and 30% most mobile atoms (red) provides the structural basis for the differential RMSDs. (F) Structures of Simulation 2 colored according to the RMSF of each residue relative to the initial structure after alignment. All these plots and figures can be obtained from the output of MDLovoFit.

Mentions: The standard CαRMSDs of the lipase in the two simulations are represented in Fig. 2A, computed relative to the first frames of the simulations. In Simulation 1 (black lines), the CαRMSD is below 1Å for most of the simulation. The protein structure, therefore, is very stable, and does not diverge from the initial structure. In Simulation 2 (red lines) the standard RMSD (Fig. 2A) increases much more and becomes greater than 3Å at about 25 ns. Clearly, the structure undergoes some structural change.


Automatic identification of mobile and rigid substructures in molecular dynamics simulations and fractional structural fluctuation analysis.

Martínez L - PLoS ONE (2015)

Analysis of the mobility of two simulations of a Burkholderia cepacia lipase [14] in mixtures of water and sorbitol.Different RMSD profiles are observed: (A) The standard RMSD of Simulation 1 (black) is much lower than the RMSD of Simulation 2 (red). (B) RMSD as a function of the fraction of the atoms considered in the alignment. These plots indicate that in Simulation 2 (red), there is a subset of about 25 to 30% of the atoms which are responsible for the greater overall RMSD observed in panel (A). (C) In both simulations, 70% of the atoms can be superposed to less than 1Å (RMSDL—dotted lines, black for Simulation 1, red for Simulation 2). The remaining 30% of the atoms behave differently in each simulation (RMSDH—solid lines, same colors). (D) and (E) Superposition of the frames and coloring of the 70% least mobile atoms (blue) and 30% most mobile atoms (red) provides the structural basis for the differential RMSDs. (F) Structures of Simulation 2 colored according to the RMSF of each residue relative to the initial structure after alignment. All these plots and figures can be obtained from the output of MDLovoFit.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4376797&req=5

pone.0119264.g002: Analysis of the mobility of two simulations of a Burkholderia cepacia lipase [14] in mixtures of water and sorbitol.Different RMSD profiles are observed: (A) The standard RMSD of Simulation 1 (black) is much lower than the RMSD of Simulation 2 (red). (B) RMSD as a function of the fraction of the atoms considered in the alignment. These plots indicate that in Simulation 2 (red), there is a subset of about 25 to 30% of the atoms which are responsible for the greater overall RMSD observed in panel (A). (C) In both simulations, 70% of the atoms can be superposed to less than 1Å (RMSDL—dotted lines, black for Simulation 1, red for Simulation 2). The remaining 30% of the atoms behave differently in each simulation (RMSDH—solid lines, same colors). (D) and (E) Superposition of the frames and coloring of the 70% least mobile atoms (blue) and 30% most mobile atoms (red) provides the structural basis for the differential RMSDs. (F) Structures of Simulation 2 colored according to the RMSF of each residue relative to the initial structure after alignment. All these plots and figures can be obtained from the output of MDLovoFit.
Mentions: The standard CαRMSDs of the lipase in the two simulations are represented in Fig. 2A, computed relative to the first frames of the simulations. In Simulation 1 (black lines), the CαRMSD is below 1Å for most of the simulation. The protein structure, therefore, is very stable, and does not diverge from the initial structure. In Simulation 2 (red lines) the standard RMSD (Fig. 2A) increases much more and becomes greater than 3Å at about 25 ns. Clearly, the structure undergoes some structural change.

Bottom Line: These substructures are automatically identified by the method.The algorithm consists of the iterative superposition of the fraction of structure displaying the smallest displacements.Examples are given to illustrate the interpretative advantages of this strategy.

View Article: PubMed Central - PubMed

Affiliation: Institute of Chemistry, University of Campinas-UNICAMP, Campinas, SP, Brazil.

ABSTRACT
The analysis of structural mobility in molecular dynamics plays a key role in data interpretation, particularly in the simulation of biomolecules. The most common mobility measures computed from simulations are the Root Mean Square Deviation (RMSD) and Root Mean Square Fluctuations (RMSF) of the structures. These are computed after the alignment of atomic coordinates in each trajectory step to a reference structure. This rigid-body alignment is not robust, in the sense that if a small portion of the structure is highly mobile, the RMSD and RMSF increase for all atoms, resulting possibly in poor quantification of the structural fluctuations and, often, to overlooking important fluctuations associated to biological function. The motivation of this work is to provide a robust measure of structural mobility that is practical, and easy to interpret. We propose a Low-Order-Value-Optimization (LOVO) strategy for the robust alignment of the least mobile substructures in a simulation. These substructures are automatically identified by the method. The algorithm consists of the iterative superposition of the fraction of structure displaying the smallest displacements. Therefore, the least mobile substructures are identified, providing a clearer picture of the overall structural fluctuations. Examples are given to illustrate the interpretative advantages of this strategy. The software for performing the alignments was named MDLovoFit and it is available as free-software at: http://leandro.iqm.unicamp.br/mdlovofit.

No MeSH data available.


Related in: MedlinePlus