Embed Size (px)
Citation preview
Transforming drug discovery and materials research
Accelerating Drug Discovery with Computational Chemistry
Graduate Student Guest Lecture Mikolai Fajer, PhD
Senior Scientist
Transforming drug discovery and materials research
Drug Discovery
Designing Drugs is Hard
• Need to simultaneously optimize many properties: – Bioavailability – Clearance / half-life – Permeability
– CYP inhibition – hERG blockade – Synthesizability
– Potency – Selectivity – Solubility
– Toxicity
3
4
Designing Drugs is Hard
• Need to simultaneously optimize many properties: – Bioavailability – Clearance / half-life – Permeability
– CYP inhibition – hERG blockade – Synthesizability
– Potency – Selectivity – Solubility
– Toxicity
Mol 4
Mol 3
Mol 2
Mol 1Pr
op
erty
1
Pro
per
ty 2
Pro
per
ty 3
Pro
per
ty 4
Pro
per
ty 5
Pro
per
ty 6
Pro
per
ty 7
Pro
per
ty 8
Pro
per
ty 9
Pro
per
ty 1
0
Designing Drugs is Hard
• Need to simultaneously optimize many properties: – Bioavailability – Clearance / half-life – Permeability
– CYP inhibition – hERG blockade – Synthesizability
– Potency – Selectivity – Solubility
– Toxicity
Mol 4
Mol 3
Mol 2
Mol 1Pr
op
erty
1
Pro
per
ty 2
Pro
per
ty 3
Pro
per
ty 4
Pro
per
ty 5
Pro
per
ty 6
Pro
per
ty 7
Pro
per
ty 8
Pro
per
ty 9
Pro
per
ty 1
0Statistical Thermodynamics
5
6
Even Late in a Project, Activity Cliffs are a Problem
4
5
6
7
8
9
10
11
1 Time (2+ Years / 1,390 cmpds)
Com
poun
d Po
tenc
y (lo
g un
its)
Representative Pre-FEP+ Project (2010) (12% of molecules tight binding)
• ADMET tuning repeatedly leads to losses of potency even in late in project
• Primary series liabilities may force the project team to effectively start over – Core-related toxicity – Competitor IP filings
• Earlier SAR may not transfer to backup series
Most Drug Discovery Projects Fail to Deliver a Drug into the Clinic
Data extracted from: Nature Review Drug Discovery, 2016, 14, 475; Nature Review Drug Discovery, 2014, 13, 419; Nature Review Drug Discovery 2010, 9, 203.
34% Succeed Send molecule into the clinic
22%
44% 34%
22% Fail Due to Poor Initial Target Selection Target engagement not efficacious for treating disease; on-target toxicity; corporate portfolio rebalancing
44% Fail Due to Ligands of Insufficient Quality Ligand toxicity problems;
efficacy not observed at achievable dose; poor PK/PD
7
Current State of the Art
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
2,000 design ideas synthesized
10% probability the designed compound is potent enough
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
200 potent synthesized compounds
0.41% probability a potent compound meets other project criteria
34% chance of drug entering clinic
Trial-and-error
Deduced from: Nature Review Drug Discovery, 2016, 14, 475; Nature Review Drug Discovery, 2014, 13, 419; Nature Review Drug Discovery 2010, 9, 203.
8
Transforming drug discovery and materials research
Free Energy Perturbation
10
How to Rigorously Compute Affinity – Compute All the Terms
Molecule in water
Molecule in bound conformation
Desolvated molecule
Protein in water
Protein in bound conformation
Desolvated protein
Molecule bound to protein
∆G(5)
∆G(1) ∆G(3)
∆G(2) ∆G(4)
∆Gbind = ∆G(1) + ∆G(2) + ∆G(3)
+ ∆G(4) + ∆G(5)
11
FEP Provides a Rigorous Method to Compute Relative Binding Affinity
Free Energy Perturbation (FEP) technology
• Rigorous calculation of the binding affinity difference between two ligands – Series of molecular dynamics
simulations are run where Ligand A is alchemically transformed into Ligand B
– Appropriate Stat. Mech. is used to rigorously compute the binding free energy difference between Ligands A and B
FEP and the Abstract λ-coordinate
12
0V 1V)(λV
0=λ 1=λ5.0=λ01G
00G 0GΔ
ΔG0 =G10 −G0
0 = −kT ln exp −ΔE kT( )
Real starting state Real final state Alchemical intermediates
(BAR or MBAR used in practice)
• What do the various bits mean?
– Forcefield
– Forcefield, Perturbation Size
– Sampling
• Thermodynamic Integration
• Free Energy Perturbation
• Expectation Value
Dissecting Expectation Values
13
∆𝐺=−𝑘𝑇ln�⟨exp[−∆𝑉⁄𝑘𝑇 ]⟩
∆𝐺=∫0↑1▒⟨𝜕𝑉/𝜕𝜆 ⟩↓𝜆 𝑑𝜆
⟨𝕆⟩= ∫↑▒𝕆(𝑋)𝑒↑−𝑉(𝑋) 𝑑𝑋 /∫↑▒𝑒↑−𝑉(𝑋) 𝑑𝑋
∫↑▒⋯𝑑𝑋
𝑉(𝑋)
𝕆(𝑋) exp[−∆𝑉⁄𝑘𝑇 ] 𝜕𝑉/𝜕𝜆
∫↑▒𝑒↑−𝑉(𝑋) 𝑑𝑋
14
1. Predictions must be accurate 2. Calculation must take much less time
than compound synthesis 3. Calculation setup must be
straightforward
Schrödinger’s Approach: FEP+
Secure Cloud Platform Invested heavily in internal GPU cluster and technology for running on the Cloud
Schrodinger FEP+ Advances
High Accuracy Across Chemical Space
High Throughput: Fast and efficient calculations
Built an accurate force field (molecular energy function and interatomic potentials) that covers effectively all of relevant chemical space.
High Throughput: Speed & Efficiency Deployed GPU-enabled molecular dynamics engine - ~50-100s performance advantage over CPUs. Invested in technology for running securely on the cloud. Extensive use of enhanced sampling methods.
Requirements for FEP calculations to enable drug discovery
Automated Setup Tools Built automation tools to make jobs significantly easier to run and troubleshoot
~ 200 person-years of effort to make advances in the science
Transforming drug discovery and materials research
Performance and Validation
Head-to-Head Study: FEP+ vs. Traditional Methods for Compound Prioritization
Source: Kuhn et al. J. Med. Chem. 2017, 60(6):2485-2497.
Study Overview
Methods
Prospective Evaluation of Free Energy Calculations for thePrioritization of Cathepsin L InhibitorsBernd Kuhn,†,# Michal Tichy,‡,# Lingle Wang,§,# Shaughnessy Robinson,§ Rainer E. Martin,†
Andreas Kuglstatter,† Jorg Benz,† Maude Giroud,‡ Tanja Schirmeister,∥ Robert Abel,*,§Francois Diederich,*,‡ and Jero me Hert*,††Roche Pharmaceutical Research and Early Development (pRED), Roche Innovation Center Basel, F. Hoffmann-La Roche Ltd.,Grenzacherstrasse 124, 4070 Basel, Switzerland‡Laboratorium fur Organische Chemie, ETH Zurich, Vladimir-Prelog-Weg 3, 8093 Zurich, Switzerland§Schrodinger, Inc., 120 West 45th Street, New York, New York 10036, United States∥Institut fur Pharmazie und Biochemie, Johannes Gutenberg-Universitat Mainz, Staudinger Weg 5, 55128 Mainz, Germany
*S Supporting Information
ABSTRACT: Improving the binding affinity of a chemical series by systematically probing one of its exit vectors is a medicinalchemistry activity that can benefit from molecular modeling input. Herein, we compare the effectiveness of four approaches inprioritizing building blocks with better potency: selection by a medicinal chemist, manual modeling, docking followed by manualfiltering, and free energy calculations (FEP). Our study focused on identifying novel substituents for the apolar S2 pocket ofcathepsin L and was conducted entirely in a prospective manner with synthesis and activity determination of 36 novelcompounds. We found that FEP selected compounds with improved affinity for 8 out of 10 picks compared to 1 out of 10 for theother approaches. From this result and other additional analyses, we conclude that FEP can be a useful approach to guide thistype of medicinal chemistry optimization once it has been validated for the system under consideration.
■ INTRODUCTIONFree energy calculation approaches, such as free energyperturbation (FEP), have been around for a long time1−5 buthad only limited impact in the drug discovery process so far.Likely reasons for their historically restrained use includelengthy simulation times not practical in fast-paced projectenvironments combined with overstated accuracy levels basedon small test set retrospective analyses which did not translatewhen employed prospectively in real-world systems. FEP hasnow taken advantage of improved sampling algorithms6,7 andforce-field quality8 and is profiting from the increasedavailability of low-cost parallel computing. Speed and accuracyappear to have progressed significantly.7,9,10 This has in turn ledto recent accounts of successful industrial applications of FEPin active drug discovery projects.11−13
Here, we investigate the application of FEP in a typical drugdiscovery use case where the goal is to prioritize compounds forsynthesis. One way for therapeutic project teams to furtherexplore the structure−activity relationship (SAR) of a hit series
is to engage in parallel synthesis: A common setting involves ascaffold with one or several defined exit vectors and a set ofchemical reactions with the goal of optimizing side chains. Thenumber of suitable reactants accessible (internally orpurchasable) can be very large (hundreds, thousands, ormore). The task of molecular modeling consists then inprioritizing building blocks with respect to binding affinity inorder to limit the amount of synthesis and experimental testingrequired.A prerequisite for this exercise is the availability of an initial
ligand together with structural information, an experimentalcocrystal structure describing the binding mode to the protein.We picked human cathepsin L (hCatL), a cysteine protease,which can be inhibited by ligands with an activated nitrile groupforming a covalent thioimidate adduct with the catalytic Cys25.Previous SAR and structural studies with aryl nitriles (Figure 1)
Received: December 28, 2016
Article
pubs.acs.org/jmc
© XXXX American Chemical Society A DOI: 10.1021/acs.jmedchem.6b01881J. Med. Chem. XXXX, XXX, XXX−XXX
• Cathepsin L is a lysosomal endopeptidase which plays an important role in protein degradation and apoptosis
• Good structural data available along with some SAR • The collaboration was used as a test of FEP+ scoring vs.
traditional approaches to optimize a 200 nM lead compound
• 3,325 R-group idea library provided by the collaborator
• Design ideas were docked and scored using Glide and Prime MM-GB/SA for idea triage
• 92 molecules selected for FEP+ scoring
• Head-to-head comparison: – 10 molecules prioritized by FEP+ scoring
– 10 molecules prioritized by an experienced med chemist (Med Chem)
– 10 molecules prioritized an experienced modeler using any technique other than free energy calculations (SBDD)
– 10 molecules prioritized by docking and filtering (Docking)
16
Results: FEP+ Dramatically Out-Performed in Prioritizing Potent Compounds
FEP+ Med Chem SBDD Docking
Cmpd. ID Exp Ki (nM) Cmpd. ID Exp Ki (nM) Cmpd. ID Exp Ki (nM) Cmpd. ID Exp Ki (nM)
3 12 3 12 3 12 22 77
37 25 11 279 14 217 29 304
31 27 9 515 16 505 28 358
33 30 7 952 13 1010 26 411
35 77 6 1800 20 1020 23 671
34 91 4 3020 17 2790 13 1010
30 123 8 (cis) 3500 15 3860 20 1020
38 167 5 >5100 18 >5100 24 5100
36 1430 8 (trans) >5100 19 >5100 25 >5100 32 1750 10 >5100 21 >5100 27 >5100
8/10 tighter binding than the reference
1/10 tight binding than the reference
1/10 tight binding than the reference
1/10 tight binding than the reference
Note: Molecule 3 was very close to the known SAR, tight binding was unsurprising 17
18
FEP+ Selected Molecules were Diverse and Manifested Bond Topologies not Previously Known to Bind Cathepsin L
Expt. Ki = 123 nM Pred. Ki = 11 nM
Expt. Ki = 27 nM Pred. Ki = 15 nM
Expt. Ki = 12 nM Pred. Ki = 18 nM
Expt. Ki = 1750 nM Pred. Ki = 15 nM
Expt. Ki = 30 nM Pred. Ki = 16 nM
Expt. Ki = 25 nM Pred. Ki = 24 nM
Expt. Ki = 91 nM Pred. Ki = 103 nM
Expt. Ki = 77 nM Pred. Ki = 19 nM
Expt. Ki = 1430 nM Pred. Ki = 18 nM
Core
37 31 3 32
33 34 35 30 36
FEP+ Domain of Applicability
1. At least one high quality crystal structure with cocrystalized series ligand
2. Reasonable expectation of a conserved binding mode across the series
3. Minimal tautomeric, ionization-state, and stereochemistry uncertainties across
the series
4. High reliability experimental binding data from the same assay for all compounds
5. Assay data and crystal structures are for the same protein construct
19
FEP+ Domain of Applicability
• Failing to meet these criteria doesn’t necessarily guarantee failure, but it does make it more likely
• In general, the cleanliness of the experimental data is much more important than the identity of the target or the type of ligand modification
20
• Accuracy: beginning to approach experimental error for amenable targets
• Speed: ~50X faster per molecule – 6 hours/compound vs. ~3 weeks/compound
• Throughput: ~250X higher – e.g., 5000 molecules in 1 week vs. 5 years
• Project synthesis resources can be focused where they will be most productive
FEP+ as a ’Computational Assay’
21
How Can a Rigorous Computational Binding Assay Impact Preclinical Drug Discovery Projects?
• Faster potency optimization with fewer synthesized compounds à improve efficiency of MedChem cycles
• Ability to de-risk challenging chemical synthesis and affordably explore a greater diversity of chemical space
• Better maintain potency while simultaneously tuning ADMET properties during lead optimization: – Binding selectivity – Mutational resistance – Solubility – Membrane permeability
1
2
3
22
FEP+ Provides Unprecedented Control of Compound Affinity
4
5
6
7
8
9
10
11
1 Time (2+ Years / 1,390 cmpds)
Com
poun
d Po
tenc
y (lo
g un
its)
Representative Pre-FEP+ Project (2010) (12% of molecules tight binding)
4
5
6
7
8
9
10
11
0 Time (11 Months, 509 cmpds)
Com
poun
d Po
tenc
y (lo
g un
its)
Post-FEP+ Project (2016) (66% of molecules tight binding)
23
FEP+ was used to simultaneously optimize ligand binding potency, binding selectivity, and aqueous solubility
FEP+ pKi=11
FEP+ pKi=8
FEP+ Selec > 1000x
FEP+ Sol > 100 uM
FEP+ Selec < 0.1x
FEP+ Sol < 0.5 uM
pKi > 9 > 100x selectivity > 20 uM solubility
This approach has allowed us rapidly identify multiple highly potent, selective, and soluble ligands to advance the project
24
• FEP+ scoring correctly predicted molecule D would be much more selective and soluble than molecule F
• Traditional Med. Chem. design strategies have great difficult anticipating such trends
FEP+ scoring accurately captures highly counter intuitive potency, selectivity, and solubility trends
Core Core
Mol. D: pKi > 9 Selec. < 100x Solub. < 10 µM
Mol. E: pKi > 9 Selec. > 100x Solub. > 20 µM
25
26
Drug Discovery Guided by a Rigorous Computational Method
2,000 Pool of design ideas, ranked by FEP+ scoring each med chem cycle; best 2,000 synthesized over the project
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
>>200potent synthesized compounds
0.41%probability a potent compound meets other project criteria
>>34% chance of drug entering clinic
>>10%probability the designed compound is potent enough
FEP+ scoring
27
Drug Discovery Guided by a Rigorous Computational Method
>>2,000 Pool of design ideas, ranked by FEP+ scoring each med chem cycle; best 2,000 synthesized over the project
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
O
N
ONNNH
NO
O
N
ONNNH
O
N
O
N
ONNNH
O
O O
O
O
N
NNNH
O
O N
ONNNH
Cl
O
N
ONNNH
OH
OH
>>200potent synthesized compounds
0.41%probability a potent compound meets other project criteria
>>34% chance of drug entering clinic
>>10%probability the designed compound is potent enough
FEP+ scoring
Transforming drug discovery and materials research
In the Trenches
• BACE1 cleaves amyloid precursor protein yielding the β-amyloid peptide that aggregates into plaques in Alzheimer’s disease
• We have: – A crystal structure with a reference
compound – A suggestion that changing the size of
the ring in the P2’ pocket could improve results
– Enumeration of the other end of the molecule for further affinity tuning
– Knowledge that the protein flap and 10s loop are flexible
The Problem
Acylguanidine Beta Secretase 1 Inhibitors: A Combined Experimentaland Free Energy Perturbation StudyHenrik Keranen,† Laura Perez-Benito,‡,§ Myriam Ciordia,⊥ Francisca Delgado,⊥ Thomas B. Steinbrecher,∥
Daniel Oehlrich,# Herman W. T. van Vlijmen,† Andres A. Trabanco,⊥ and Gary Tresadern*,§
†Computational Chemistry, Janssen Research & Development, Janssen Pharmaceutica N. V., Turnhoutseweg 30, B-2340 Beerse,Belgium‡Laboratori de Medicina Computacional Unitat de Bioestadistica, Facultat de Medicina, Universitat Autonoma de Barcelona, 08193,Bellaterra, Spain§Computational Chemistry, Janssen Research and Development, Janssen-Cilag, c/ Jarama 75A, 45007, Toledo, Spain⊥Neuroscience Medicinal Chemistry, Janssen Research and Development, Janssen-Cilag, c/ Jarama 75A, 45007, Toledo, Spain∥Schrodinger GmbH, Dynamostrasse 13, 68165 Mannheim, Baden-Wurttemberg, Germany,#Neuroscience Medicinal Chemistry, Janssen Research & Development, Janssen Pharmaceutica N. V., Turnhoutseweg 30, B-2340Beerse, Belgium
*S Supporting Information
ABSTRACT: A series of acylguanidine beta secretase 1 (BACE1) inhibitors withmodified scaffold and P3 pocket substituent was synthesized and studied with freeenergy perturbation (FEP) calculations. The resulting molecules showed potencies inenzymatic BACE1 inhibition assays up to 1 nM. The correlation between the predictedactivity from the FEP calculations and the experimental activity was good for the P3pocket substituents. The average mean unsigned error (MUE) between prediction andexperiment was 0.68 ± 0.17 kcal/mol for the default 5 ns lambda window simulationtime improving to 0.35 ± 0.13 kcal/mol for 40 ns. FEP calculations for the P2′ pocketsubstituents on the same acylguanidine scaffold also showed good agreement withexperiment and the results remained stable with repeated simulations and increased simulation time. It proved more difficult touse FEP calculations to study the scaffold modification from increasing 5 to 6 and 7 membered-rings. Although prediction andexperiment were in agreement for short 2 ns simulations, as the simulation time increased the results diverged. This wasimproved by the use of a newly developed “Core Hopping FEP+” approach, which also showed improved stability in repeatcalculations. The origins of these differences along with the value of repeat and longer simulation times are discussed. This workprovides a further example of the use of FEP as a computational tool for molecular design.
■ INTRODUCTION
The accurate prediction of protein−ligand binding affinities isof major interest.1 Rigorous approaches can calculate thebinding free energy difference between two structurally similarligands by making use of alchemical structural modifications.Free-energy perturbation (FEP) or thermodynamic integration(TI), using molecular dynamics (MD) or Monte Carlosimulations, are among the widely used approaches.2 Thesemethods are ideal for drug discovery lead optimization whereclose structural analogues are synthesized and their propertiescompared to previous best leads. Computation of accuraterelative binding affinities can make a big impact in this costlyphase. Also, it avoids the computationally expensive predictionof absolute binding free energies. Calculating relative protein−ligand binding affinities in this way dates back at least thirtyyears.3−8 Lately, new sampling algorithms, improved forcefields and low-cost parallel computing (often graphicsprocessing units GPU), have improved accuracy and turn-around time.9−11 Reports of large scale and industrial
applications are emerging,12−18 including work from ourlaboratories investigating FEP applied in lead optimization.19,20
Here, we explore FEP predictions of the binding energies ofβ-secretase 1 (BACE1) inhibitors in the different subpockets ofthe active site. Cleavage of amyloid precursor protein by thisaspartyl protease leads to increases of β-amyloid (Aβ) peptidesthat aggregate forming senile plaques, one of the neuro-pathological features in Alzheimer’s disease (AD).21 Contem-porary BACE1 inhibitors contain an amidine/guanidine moietywithin a heterocycle of varying size.22 This protonated groupforms an optimal hydrogen-bonding network to the catalyticaspartate dyad, Figure 1. Also, a quaternary alpha sp3 carbonprovides an ideal vector to fill the P1−P3 and P2′ pockets ofthe binding site (Figures 1 and 2).23,24 These molecules haveimproved drug like properties and multiple examples are inclinical trials.25 Given the huge pharmaceutical interest infinding new treatments for AD, BACE1 is a very well-studied
Received: November 22, 2016Published: January 19, 2017
Article
pubs.acs.org/JCTC
© 2017 American Chemical Society 1439 DOI: 10.1021/acs.jctc.6b01141J. Chem. Theory Comput. 2017, 13, 1439−1453
29
• Change the R-groups • Change the size of the ring
Chemical Space
30
Perturbation Size
Testing the Ring Size
31
R N Compound ΔΔG @ 2ns (kcal/mol)
ΔΔG @ 5ns (kcal/mol)
ΔΔG @ 10ns (kcal/mol)
ΔΔG @ 40ns (kcal/mol)
0→1 8a→17a -5.2±2.6 -2.0±2.3 -1.5±0.7 -2.3±3.3
0→2 8a→27a -1.3±1.8 1.1±1.5 -0.2±1.9 1.9±2.6
0→1 8c→17c -3.0±1.3 -5.0±2.5 -2.3±1.7 -3.7±1.6
0→2 8c→27c -2.1±0.5 -0.9±2.5 -1.2±1.6 0.1±3.2
0→1 8g→17g -4.6±3.0 -4.4±1.3 -1.8±0.9 -3.3±1.3
0→2 8g→27g -1.8±0.9 0.7±1.2 2.2±1.8 0.9±1.9
-3.2±0.3
-1.3±0.3
-3.5±0.1
-1.8±0.1
-1.9±0.1
-0.9±0.1
Next
What causes large deviations between runs?
32
Compd. ΔΔG @ 2ns (kcal/mol)
ΔΔG @ 40ns (kcal/mol)
8c→27c -2.1±0.5 0.1±3.2
Back
Testing the R-group enumeration
33
R N Compound 2ns 5ns 10ns 40ns
1 17a -10.8±0.1 -11.6±0.3 -11.4±0.3 -11.7±0.1
1 17c -11.7±0.1 -11.8±0.2 -12.3±0.1 -12.1±0.1
1 17g -14.2±0.4 -13.9±0.3 -13.5±0.3 -13.7±0.1
-11.4±0.1
-12.1±0.4
-12.6±0.2
Back
Minimize the Perturbation
34
Full Ring Perturbation
“Core Hopping” Perturbation
“Core Hopping”
• What have we learned for the next round?
• Quantitative accuracy?
• Predictive accuracy?
Evaluation
35
Compound ΔΔG @ 40ns (kcal/mol)
ΔΔG exp. (kcal/mol)
8a→17a -2.3±3.3 -3.2±0.3
8a→27a 1.9±2.6 -1.3±0.3
![arXiv:1712.06113v3 [physics.chem-ph] 22 Mar 2018 · Accelerating the discovery of molecules and materi- als with desired properties is a long-standing challenge in computational chemistry](https://img.dokumen.tips/doc/110x75/5d4a4b5e88c9931a7c8b57f9/arxiv171206113v3-22-mar-2018-accelerating-the-discovery-of-molecules-and.jpg)
