"Validating Water Placement: Experimental and Computational Methods" was presented by Greg Warren, Ph.D., Senior Applications Scientist on Wednesday, August 9, 2017 at 12pm ET / 9am PT (US).
There are a number of computational methods available for assessing the structure and energetics of water as it solvates macromolecular structures. While there has been a large amount of interest in the results generated by computational methods ranging from semi-continuum (SZMAP) to explicit methods (WATERMAP) a good validation of any method’s ability to predict both the position and orientation of “bound” water has been lacking.
There are several published validations of bound water position predictions but each of these studies contains design flaws because of the experimental data used to validate the predictions namely X-ray diffraction structures. X-ray diffraction structures, no matter the resolution, suffer from an inability to consistently identify hydrogen atom positions which would provide water orientation information. An additional problem with prior studies that these studies have not carefully selected for room temperature only structures.
We present a preliminary analysis of a water prediction validation study that relies on two structures from Phanerochaete chrysosporium (PcCel45A) endoglucanase where both X-ray (0.99 Å) and neutron (1.5 Å) diffraction data was collected on a single crystal at room temperature. We will use this experimental data to assess the water position and orientation predictions of three methods: 1) semi-continuum method SZMAP, 2) an explicit method GROMACS 1 μs MD simulation with a single protein in a box of water and 3) a GROMACS 1 μs crystal lattice MD simulation where 32 copies of the protein simulate 2x2x2 unit cell lattice. In addition to standard water simulations we performed simulations where the solvent mimicked protein crystallization (mother liquor) conditions. This presentation will show examples of water orientation and placement predictions and how they match the experimental data.
Greg Warren1, Mike Word1, Christopher Bayly1, Gaetano Calabro1, David Mobley2, Michael Wall3 1OpenEye Scientific Software, Inc., 2University of California, Irvine, 3Los Alamos National Laboratory