Making alchemical free energy calculations practical for drug discovery David Mobley Even modest accuracies in calculated binding free energies can have significant benefits Hypothetical pipeline: • Medicinal chemist suggests 100 derivatives or compounds per week • Your job is to pick the top 10 to carry forward Shirts, Mobley, and Brown, “Free Energy Calculations in Structure-based Drug Design”, in “Structure Based Drug Design” by Merz et al., Cambridge University Press, 2010. See also Mobley & Klimovich, JCP 2013 Even modest accuracies in calculated binding free energies can have significant benefits Hypothetical pipeline: • Medicinal chemist suggests 100 derivatives or compounds per week Question: How many molecules do • Your job is to pick the top 10 we have to make to gain a factor of to carry forward 10 in affinity? - 0.5 kcal/mol noise: Decreases # required by 8x - 1.0 kcal/mol noise: Decreases by 5x - 2 kcal/mol noise: Decreases by 3x Shirts, Mobley, and Brown, “Free Energy Calculations in Structure-based Drug Design”, in “Structure Based Drug Design” by Merz et al., Cambridge University Press, 2010. See also Mobley & Klimovich, JCP 2013 Relative binding free energy calculations are directly relevant to lead optimization P P L o ∆G 1 L 1 1 ∆∆G ∆∆G solv site P L P o 2 ∆G L 2 2 Cl Cl S S We waO nt to automateS O relativeNH2+ free energy H2N+ O S O H2N H2N+ O NH F O NH H2N+ H2N+ H2N F N H2N N NH2 cN alculaO tN ioO nNH2s forN leaH2N d oN pO timizatNioOn H2N O HO NH O O NH N O HN O S O Br NH2+ O H O N O N N Cl N O S NH O N O H2N+ N+ F HN+ 2 NH H2N N O S O N NH HN HN 2 S 2 H O N N O S O O NH HN+ F 2 HN Cl O N H2N+ N N NH2 N NH+ NH NH2 O H2N+ H2N+ Cl O O N H2N H2N H N H2N+ O N OH N N CH2+ N N O NH2 O H2N N Cl O O N O O S NH N O O S O O NH NH O H2N O S O O NH NH S NH2 NH2+ O S NH+ NH2+ H2N+ N N Cl H2 NH2 N NH2+ O Cl N O N N H2N+ O N N F O NH Cl SO NH O H2N+ N HN OO H2N N O O S O HN OO NH2+ H2N O S NH O S NH H2N N H H NH3+ N N O OH Cl Cl N O O H2N+ N O Cl O H2N O NH O NH2 H2N+ O N O N O H N N O O O NH H2N NH S O NH O O S O N O OH H O Cl N O NH S NH2 N NH O O H O S Cl H S NH O N N N N OH O S N O O F O N O N Cl S NH O O Cl Cl S S We waO nt to automateS O relativeNH2+ free energy H2N+ O S O H2N H2N+ O NH F O NH H2N+ H2N+ H2N F N H2N N NH2 cN alculaO tN ioO nNH2s forN leaH2N d oN pO timizatNioOn H2N O HO NH O O NH N O HN O S O Br NH2+ O H O N O N N Cl N O S NH O N O H2N+ N+ F HN+ 2 NH H2N N O S O N NH HN HN 2 S 2 H O N N O S O O NH HN+ F 2 HN Cl O N H2N+ N N NH2 N NH+ NH NH2 O H2N+ H2N+ Cl O O N H2N H2N One constraint limiting the moHre N H2N+ O N OH N N CH2+ N N O NH2 O H2N N Cl O O N O wNH ideOsprSead application of free energy N O O S O O NH NH O H2N O S O O NH NH S NH2 NH2+ calculations: O S NH+ NH2+ H2N+ N N Cl H2 NH2 N NH2+ O Cl N O N N H2N+ O N N F O The difficulty of setting thNH em up Cl SO NH O H2N+ N HN OO H2N N O O S O HN OO NH2+ H2N O S NH O S NH H2N N H H NH3+ N N O OH Cl Cl N O O H2N+ N O Cl O H2N O NH O NH2 H2N+ O N O N O H N N O O O NH H2N NH S O NH O O S O N O OH H O Cl N O NH S NH2 N NH O O H O S Cl H S NH O N N N N OH O S N O O F O N O N Cl S NH O O We developed a LOMAP to plan relative free energy calculations This approach has allowed larger-scale testing across ~200 compounds, and many subsequent applications Automation helps us advance our science and enable new applications Now that we have best practices, let’s automate them: Chodera’s Yank Automates alchemical free energy calculations in implicit or • explicit solvent Hydration or binding • We put it on Orion: We have a hydration floe and a draft binding • floe