Data for Machine Learning-driven Multiscale Modeling Reveals Lipid-Dependent Dynamics of RAS Signaling Proteins
RAS is a signaling protein that associates with the cell membrane; when mutated RAS is associated with 30% of human cancers. It has been proposed that RAS signaling is regulated by dynamic heterogeneity of the cell membrane. However, direct observation of this complexity requires near-atomic-scale resolution over macroscopic time- and length-scales. To bridge this disparity we construct a multiscale simulation infrastructure that uses machine learning to guide an automated workflow based on experimental data, including our newly-solved structure of active wild-type KRAS. We report distinctive patterns of local lipid composition that correlate with interfacially promiscuous RAS multimerization. These lipid “fingerprint patterns” are coupled to RAS dynamics, and are predicted to influence effector binding. The fingerprint patterns therefore may be a mechanism for regulating various cell signaling cascades.
Here we provide input parameters used in our dynamic density functional theory (DDFT) macro-scale model and in the coarse-grain Martini micro-scale simulations.
This work was performed under the auspices of the U.S. Department of Energy (DOE) by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344, Los Alamos National Laboratory (LANL) under Contract DE-AC5206NA25396, Oak Ridge National Laboratory under Contract DE-AC05-00OR22725, and FNLCR under Contract HHSN261200800001E. Release: LA-UR-19-32061, LLNL-JRNL-799963.
All data files (138 MB): RAS_data.tar
Macro-scale-model input files
cofr_ij_newmodel.txt - Macro-model parameters: Lipid-lipid direct correlations functions.
object.data - Control file for RAS motion integration (ddcMD). Contains RAS-RAS potential and state change model parameters.
1um-restart-13.i - MOOSE control file. Contains, for example, further macro model diffusion coefficients, noise parameters, timestep and accuracy controls, and pointers to initial configurations and other files model parameters (e.g., cofr_ij_newmodel.txt).
Micro-scale-simulation input files
Most of the Martini parameters files are directly from the Martini website (http://cgmartini.nl/), with modifications listed here.
KRAS-GTP-04.itp - Input parameters for RAS generated from the open-source tool martinize.py, and modified in a number of ways by LANL and LLNL (including new HVR and farnesyl parameters from Cesar Lopez - firstname.lastname@example.org). See the description in section 1.2.2 in the supplementary information.
POPX_Martini_v2.0_lipid.itp - POPC parameters with position restraints on the z-axes of the PO4 bead.
martini_v2.0_CHOL_01-fix.itp - A version of the original non-virtual-site Martini cholesterol (v01) but with updates to the cholesterol shape to reflect that of the never-virtual-sites model (v02).
martini_v2.0_CHOL_02.itp - per Martini website.
martini_v2.0_DIPE_01.itp - per Martini website.
martini_v2.0_DPSM_01.itp - per Martini website.
martini_v2.0_PAPC_01.itp - per Martini website.
martini_v2.0_PAPS_01.itp - per Martini website.
martini_v2.0_POPC_02.itp - per Martini website.
martini_v2.0_POPE_02.itp - per Martini website.
martini_v2.0_PAP6_02.itp - Phosphatidylinositol 4,5-bisphosphate lipid parameters with palmitoyl and arachidonoyl tails. Generated using the Martini lipid .itp generator (v6). The PIP headgroup parameters are updated from the original Lopez et al. 2013 JCTC parameters by Carlos Ramirez Palacios (University of Groningen).
martini_v2.0_ions.itp - Martini ion parameter file.
martini_v2.1-dna.itp - Martini 2.1 (and 2.2) general parameter file including DNA extension.
martini_v2.x_new-rf-prod.mdp - A lightly modified Martini simulation input file.
Example of a resulting single RAS micro-scale simulation (pfpatch_000001395662)
system.top - GROMACS topology file containing the list of all particles in this simulation
equilibrated-system.gro - Initial GROMACS particle coordinate file for this simulation (after initial system setup and equilibrium)
system1.9us.xtc - (135.9 MB) GROMACS trajectory file, containing snapshots of the simulation saved at 10 ns intervals until the simulation end at 1.9 microseconds.