copolymeric membranes [38]. They found that the bilayer elastic parameters can be
described at an almost quantitative level by an appropriate combination of monolayer elastic parameters.
2.2 Coarse-Grained Lipid Models
The multitude of length and time scales that matter for biophysical membrane
processes is mirrored in a wide spectrum of computational models that have been
devised to capture these scales. These range from all-atom simulations [39–43] up
to dynamically triangulated surfaces [44–47] and continuum models [48, 49]. The
region in between is becoming increasingly populated by a wealth of different
coarse-grained (commonly abbreviated “CG”) models, which capture different
aspects of a very complex physical situation, and a number of excellent reviews
exist that provide a guide to the literature [50–57].
Besides their chosen level of resolution, CG models can also be classified by the
“spirit” in which they approach a physical situation: If the focus lies on generic
mechanisms that are thought to be quite universal in their reach, there is no need to
construct models that faithfully relate to every aspect of some particular lipid.
Instead, one creates “top-down” models based on the presumed principles underlying the generic mechanisms of interest. For instance, if one wishes to understand
how a bilayer membrane interacts with a colloidal particle that is much bigger than
the thickness of the membrane, relevant aspects of the situation will likely include
the fluid curvature–elastic response of bilayer lipid membranes, but probably not
the hydrogen bonding abilities of a phosphatidylethanol head group. If, in contrast,
one wishes to understand how mesoscopic membrane properties emerge from
specific properties of their microscopic constituents, the aim is instead to construct
“bottom-up” models whose key design parameters follow in a systematic way from
those of a more finely resolved model. For instance, if one wishes to understand
how those hydrogen bonding abilities of a phosphatidylethanol head group
impact the mesoscopic phase behavior of mixed bilayers, it will not do to simply
guess a convenient head group interaction potential, even if it is eminently plausible. The latter philosophy goes under various names, such as “systematic coarse
graining” or “multiscaling” and again excellent literature and resources exist that
cover this field [58–72].
The top-down and bottom-up approaches are not necessarily mutually exclusive.
It is conceivable that certain aspects of the science are systematically matched,
while others are accounted for in a generic way by using intuition from physics,
chemistry, mathematics, or other pertinent background knowledge. Conversely,
this also means that what any given model can qualitatively or quantitatively
predict depends greatly on the way in which it has been designed; there is no
universally applicable CG model. Stated differently, systematically coarse-grained
models will not be accurate in every prediction they make, and generic models can
Computational Studies of Biomembrane Systems: Theoretical Considerations. . .
241
described at an almost quantitative level by an appropriate combination of monolayer elastic parameters.
2.2 Coarse-Grained Lipid Models
The multitude of length and time scales that matter for biophysical membrane
processes is mirrored in a wide spectrum of computational models that have been
devised to capture these scales. These range from all-atom simulations [39–43] up
to dynamically triangulated surfaces [44–47] and continuum models [48, 49]. The
region in between is becoming increasingly populated by a wealth of different
coarse-grained (commonly abbreviated “CG”) models, which capture different
aspects of a very complex physical situation, and a number of excellent reviews
exist that provide a guide to the literature [50–57].
Besides their chosen level of resolution, CG models can also be classified by the
“spirit” in which they approach a physical situation: If the focus lies on generic
mechanisms that are thought to be quite universal in their reach, there is no need to
construct models that faithfully relate to every aspect of some particular lipid.
Instead, one creates “top-down” models based on the presumed principles underlying the generic mechanisms of interest. For instance, if one wishes to understand
how a bilayer membrane interacts with a colloidal particle that is much bigger than
the thickness of the membrane, relevant aspects of the situation will likely include
the fluid curvature–elastic response of bilayer lipid membranes, but probably not
the hydrogen bonding abilities of a phosphatidylethanol head group. If, in contrast,
one wishes to understand how mesoscopic membrane properties emerge from
specific properties of their microscopic constituents, the aim is instead to construct
“bottom-up” models whose key design parameters follow in a systematic way from
those of a more finely resolved model. For instance, if one wishes to understand
how those hydrogen bonding abilities of a phosphatidylethanol head group
impact the mesoscopic phase behavior of mixed bilayers, it will not do to simply
guess a convenient head group interaction potential, even if it is eminently plausible. The latter philosophy goes under various names, such as “systematic coarse
graining” or “multiscaling” and again excellent literature and resources exist that
cover this field [58–72].
The top-down and bottom-up approaches are not necessarily mutually exclusive.
It is conceivable that certain aspects of the science are systematically matched,
while others are accounted for in a generic way by using intuition from physics,
chemistry, mathematics, or other pertinent background knowledge. Conversely,
this also means that what any given model can qualitatively or quantitatively
predict depends greatly on the way in which it has been designed; there is no
universally applicable CG model. Stated differently, systematically coarse-grained
models will not be accurate in every prediction they make, and generic models can
Computational Studies of Biomembrane Systems: Theoretical Considerations. . .
241