期刊名称：Lecture Notes in Engineering and Computer Science
印刷版ISSN：2078-0958
电子版ISSN：2078-0966
出版年度：2018
卷号：2233&2234
页码：72-77
出版社：Newswood and International Association of Engineers
摘要：The native structure of a protein corresponds
to the global minimum of free energy. The replica-exchange
method (REM) has recently been used to search for the energy
minimum in a wide range of protein conformations. For large
systems, however, applying REM can be costly because the number
of replicas required for conformational sampling increases.
In the study reported here, a variant of multidimensional REM
with automatic parameter tuning was developed. The proposed
REM enables us to reduce the number of replicas needed for the
simulation, using the fragment-based expression for the Hamiltonian.
In our fragment-based REM, we defined the residue
fragments and introduced a replica-exchange process based on
the potential energy of the “target fragment”. The number
of replicas needed in the fragment-based REM depends solely
on the number of degrees of freedom in the target fragment
rather than on the size of the whole system, so the shortcoming
of the conventional REM is overcome. Although there is a oneto-one
correspondence between replicas and temperatures in
the conventional REM, the fragment-based REM treats the
fragment size as a second parameter characterizing the replicas.
Replica groups with a small fragment size accelerate the local
structure refinement, and those with a large fragment size
accelerate the global structure refinement. In this study, an
automatic tuning scheme, which provides appropriate fragment
size adaptively, has also been developed. Computing equilibrium
distributions for peptides, we found that the proposed REM
successfully provides appropriate fragment length and thus
good conformational sampling performance.
