Sequential data assimilation for single-molecule FRET photon-counting data
-
- Yasuhiro Matsunaga
- Advanced Institute for Computational Science, RIKEN 1 , 7-1-26 Minatojima-minamimachi, Chuo-ku, Kobe, Hyogo 650-0047, Japan
-
- Akinori Kidera
- Yokohama City University 2 Graduate School of Medical Life Science, , 1-7-29 Suehiro-cho, Tsurumi, Yokohama 230-0045, Japan
-
- Yuji Sugita
- Advanced Institute for Computational Science, RIKEN 1 , 7-1-26 Minatojima-minamimachi, Chuo-ku, Kobe, Hyogo 650-0047, Japan
Description
<jats:p>Data assimilation is a statistical method designed to improve the quality of numerical simulations in combination with real observations. Here, we develop a sequential data assimilation method that incorporates one-dimensional time-series data of smFRET (single-molecule Förster resonance energy transfer) photon-counting into conformational ensembles of biomolecules derived from “replicated” molecular dynamics (MD) simulations. A particle filter using a large number of “replicated” MD simulations with a likelihood function for smFRET photon-counting data is employed to screen the conformational ensembles that match the experimental data. We examine the performance of the method using emulated smFRET data and coarse-grained (CG) MD simulations of a dye-labeled polyproline-20. The method estimates the dynamics of the end-to-end distance from smFRET data as well as revealing that of latent conformational variables. The particle filter is also able to correct model parameter dependence in CG MD simulations. We discuss the applicability of the method to real experimental data for conformational dynamics of biomolecules.</jats:p>
Journal
-
- The Journal of Chemical Physics
-
The Journal of Chemical Physics 142 (21), 214115-, 2015-06-05
AIP Publishing