You are here

Framework for Biomolecular Optimization and Design Using a New Poisson-Boltzmann Description

Award Information
Agency: Department of Health and Human Services
Branch: National Institutes of Health
Contract: 1R43GM131549-01
Agency Tracking Number: R43GM131549
Amount: $150,597.00
Phase: Phase I
Program: SBIR
Solicitation Topic Code: 400
Solicitation Number: PA18-574
Solicitation Year: 2018
Award Year: 2019
Award Start Date (Proposal Award Date): 2019-04-01
Award End Date (Contract End Date): 2020-06-30
Small Business Information
Trenton, NJ 08618-2302
United States
DUNS: 096857313
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 (609) 538-0444
Business Contact
Phone: (609) 538-0444
Research Institution

Rational drug design holds the potential for developing more effective drugs in shorter time frames than traditional trial and error approachesAccordinglysoftware tools are needed that automate the search process and can computationally direct the changes in ligand geometry and charge distribution to selectively bind at specified receptor sitesSuch analysis entails two componentsa module that evaluates a cost functione gbinding energy affinity specificityfor a given parameter sete gcharge distributionand an algorithm for adjusting the parameters to optimize the cost functionCurrent analysis options include generalized BornGBPoisson BoltzmannPBbased descriptions and molecular dynamicsMDGB methods lack geometric detail and MD is expensive so that the balance of modeling fidelity and computational cost in PB methods would seemingly commend them for drug design applicationsHoweverPB methods inadequately characterize the solvent near the moving vibrating molecular surfaceresulting in limited agreement with MD and experimental dataexhibit numerical difficultiese gsingular solutionsdiscontinuous forcesand are overly dependent on choice of surface definition thus limiting predictive reliabilityThe proposed effort has two core aimsThe first is to address the numericalcomputational and or physical modeling deficiencies of PB solversimprove agreement with MD predictionsand better match experimental data using a statistical description of the moving surface and surrounding ion distributionsAn ancillary advantage of this description is that the resulting forcesgradientsare smooth and well behavedand thus well suited for formal optimization based drug designThe second overall aim pursues this opportunity by formulating and implementing an efficient means of adjusting the design parameters to optimize binding behaviorThe approach includes an innovative and computationally efficient means of extracting the required parameter sensitivities from grid based size modified PB analysis using an adjoint variable based method to find all parameter sensitivities in one calculation with computational effort similar to solving the PB EquationA corollary benefit for both aims is implementation of the algorithms on a novel adaptive Cartesian mesh structurethat provides variable mesh spacinga singularity free representation of the solutionand fast iterative solution convergenceand has been demonstrated to lower computational resource requirements by over an order of magnitude compared to regular lattice codesThe proposed effort willDefine and incorporate smooth dielectric and ion distributionsbenchmark these distributions against MDvalidate the approach against well established small moleculebiomolecular hydration and receptor ligand binding dataandimplement a fast sensitivity analysis using adjoint variablesSuccessful development will provide a dependable and rational size modified PB based molecular design methodology thereby expediting structure based drug development and significantly improving the commercial viability of these methods NarrativeTo reduce development costs and yield more effective drugs with fewer side effectsthe drug design process is making increasing use of computational modeling methods to quantify the shapeand solvent dependent biomolecular electrostatic forcesand identify ligande gdrugconfigurations with desired binding propertiesCurrent analysis methods are not well suited to such design applications since they either lack adequate geometric and electrostatic detailGeneralized Born methodsare expensivee gexplicit solvent Molecular Dynamics simulationsparticularly in drug design where many long simulationsin the microsecond rangeare neededor are prone to discontinuous behaviorPoisson Boltzmann equation solversand one goal of the effort is to address these deficiencies and develop an intermediate modeling capability that produces continuous and smoothly varying electrostatic forcesThis analysis is combined with an innovative and efficient means of estimating parameter sensitivities whichtogether with a formal optimization processprovides a powerful computational tool for ligand optimization and the potential to revolutionize the drug design process by enabling faster throughput of more effective and specifically targeted drugs

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government