This program is used to generate coordinate files for use in molecular dynamics simulations of curved lipid bilayers
If you use our tool, please read and cite the BUMPy publication:
Kevin Boyd and Eric May. BUMPy: A Model-Independent Tool for Constructing Lipid Bilayers of Varying Curvature and Composition. Journal of Chemical Theory and Computation, 2018, 14(12), pp 6642-6652.
BUMPy is written and maintained by Kevin Boyd (kevin.boyd@uconn.edu), in the lab of Eric May at the University of Connecticut.
No installation is necessary as long as you have a python (v3) interpreter
Just numpy and scipy!
BUMPY is designed to be used at the command line, with something like the command "python bumpy.py [ options ]". All you need is the bumpy.py file! You can also take the file out of the BUMPy directory and use it wherever you want.
A novel feature of BUMPy is quantitative estimation of areas in curved bilayers using the monolayer pivotal plane - a surface within the monolayer that does not undergo area changes upon curvature deformations. On the command line, this option is controlled with the -z flag. The pivotal plane location is composition-dependent. If you do not select a pivotal plane value at the command line, one will be selected for you, though we suggest that you read the publication for an explanation of the potential consequences.
It depends on what you're using the system for. Having an inaccurate pivotal plane estimate when building these systems leads to area mismatch. We quantify some of the effects of such area mismatch in our publication. The summary is, the effects of this area mismatch are quantifiable for a number of observables such as lipid splay and diffusion, but generally quite small if your estimated pivotal plane location is within a few angstroms of the true value. This is good news, as most of the lipids we've calculated pivotal plane locations for only vary in location by a few angstroms. 10 A is a good starting guess. The key is to know what properties you want to measure, and if the subtle differences due to area mismatch are on the same order as the degree of precision you want in your measurements. Do take a look at our publication to see what we're talking about!
By mixing and matching different building block shapes, the potential arises for clashes at shape interfaces, which can (and typically does) lead to non-finite forces during energy minimization. To allow minimization to proceed, we use soft-core potentials to scale down short-range nonbonded interactions. If you are using Gromacs, the following .mdp snippet (taken from the CHARMM-GUI's suggested minimization scheme) can be used - just paste it into a typical minimization script, and minimization should work.
free-energy = yes
init-lambda = 0.01
sc-alpha = 4
sc-power = 2
sc-coul = yes
nstdhdl = 0
couple-moltype = system
; we are changing both the vdw and the charge. In the initial state, both are on
couple-lambda0 = vdw-q
; in the final state, both are off.
couple-lambda1 = none
couple-intramol = yes
You may also need to add -DFLEXIBLE
to the DEFINE
mdp field for some Martini lipids, or minimization will hang at the first step.
Please note that the use of soft-core potentials slows down minimization by about an order of magnitude. We therefore suggest a brief (~50 step) minimization using soft-core potentials, followed by a typical minimization without soft-core potentials. We have found that every system we've created in BUMPy can be successfully minimized with these techniques, so please do let us know if you come across a usage case where soft-core potentials are not sufficient!
To report bugs, email Kevin Boyd at kevin.boyd@uconn.edu
Known bugs: