Table 2.
Summary of computational methods in all SAMPL5 host-guest submissions
| Method | Conformational sampling | Energy model | Solvent model | SAMPL5 Refs. |
|---|---|---|---|---|
| OAH/OAMe | ||||
| APR-OPC (C, E) | MD; docking | GAFF/RESP | OPC | [41] |
| APR-TIP3P (C, E) | MD; docking | GAFF/RESP | TIP3P | [41] |
| BEDAM (C, I) | MD | OPLS-2005 | AGBNP2 | [54] |
| DFT/TPSS-ca (Q, I) | Manual | DFT-D3 (TPSS functional) | COSMO-RS | [55] |
| DFT/TPSS-na (Q, I) | Manual | DFT-D3 (TPSS functional) | COSMO-RS | [55] |
| DLPNO-CCSD(T)a (Q, I) | Manual | DLPNO-CCSD(T) | COSMO-RS | [55] |
| Metadynamics (C, E) | Funnel metadynamics | GAFF/RESP | TIP3P | [56] |
| MMPBSA-GAFF (C, I) | MD | GAFF/RESP | PBSA; TIP3P | [56] |
| MMPBSA-OPLSb (C, I) | MD | OPLS/RESP | PBSA; TIP3P | [56] |
| MovTyp-1 (C, I) | Mixed torsion/low mode | KECSA 1 | Implicit | [40] |
| MovTyp-2 (C, I) | Mixed torsion/low mode | KECSA 2 | Implicit | [40] |
| TI/BARc (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| TI-psc (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| TI-rawc (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| HBARd (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| HBAR-psd (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| HBAR-ps1d (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| HBAR-ps2d (C, E) | MD; docking | CGenFF | TIP3P | [57] |
| SOMD-1e (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-2 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-3 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-4 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| CBClip | ||||
| BAR-ab-initiof (C, E) | MD | CGenFF | TIP3P | [59] |
| BAR-dockf (C, E) | MD; docking | CGenFF | TIP3P | [59] |
| TI-ab-initiof (C, E) | MD | CGenFF | TIP3P | [59] |
| TI-BARf (C, E) | MD; docking | CGenFF | TIP3P | [59] |
| TI-dockf (C, E) | MD; docking | CGenFF | TIP3P | [59] |
| BEDAM (C, I) | MD | OPLS-2005 | AGBNP2 | [54] |
| MovTyp-1 (C, I) | Mixed torsion/low mode | KECSA 1 | Implicit | [40] |
| MovTyp-2 (C, I) | Mixed torsion/low mode | KECSA 2 | Implicit | [40] |
| SOMD-1e (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-2 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-3 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
| SOMD-4 (C, E) | MD | GAFF/RESP | TIP3P | [58] |
APR attach-pull-release approach [60], OPC “optimal” 3-charge, 4-point rigid water model [61]; TIP3P transferable interaction potential three-point [47]; BEDAM binding energy distribution analysis method [62]; DLPNO-CCSD(T) domain based, local pair natural orbital-coupled-cluster single double and perturbative triple excitations [63]; DFT-D3 density functional theory with the latest dispersion corrections [64]; MovTyp Movable Type method [53]; SOMD single topology relative free energy calculations performed with Sire/OpenMM6.3 software [65, 66]; BAR Bennett acceptance ratio [50]; TI thermodynamic integration [49, 67]; GAFF generalized AMBER force field [68]; CGenFF CHARMM generalized force-field [69]; RESP restrained electrostatic potential [70]; OPLS-2005 optimized potentials for liquid simulations 2005 force field [71, 72]; KECSA knowledge-based and empirical combined scoring algorithm [73]; AGBNP2 analytical generalized born plus non-polar 2 [74]; COSMO-RS conductor-like screening model for real solvents [75]; MMPBSA molecular mechanics Poisson Boltzmann/solvent accessible surface area [52]. The classifications of the energy model: quantum (Q) or classical (C), as well as the solvent model: implicit (I) or explicit (E) are listed in parentheses following the name of each method
DLPNO-CCSD(T) and DFT/TPSS-n (n indicates neutralized) used neutralized hosts, yet fully charged guests. DFT/TPSS-c (c indicates charged) used both fully charged hosts and guests
The MMPBSA-OPLS approach was only used to generate predictions for the OAMe subset
TI was used to compute binding affinities in both TI-raw and TI-ps. Lowest values of the computed binding free energies were reported in TI-raw. In TI-ps, possible corrections were added through calculating the relative pKa/pKb of the ligands to known analogs
Binding free energies in HBAR submission were computed by Hamiltonian replica exchange method (HREM) combined with the BAR method. The protonation state correction was used in HBAR-ps, HBAR-ps1 and HBAR-ps2. Results from the neutralized-only systems were reported as HBAR-ps1 and those from systems at the experimental ionic strength were reported as HBAR-ps2. TI/BAR prediction only reported binding affinities for the OAH subset based on the averaged results computed by TI and HREM/BAR for each guest
All SOMD predictions were produced based on single topology relative free energy calculations combined with multistate Bennet acceptance ratio (MBAR) method, but with different protocols. SOMD-1: No corrections; SOMD-2: includes a correction term for long-range dispersion interactions; SOMD-3: same as SOMD-2, but a correction term for the use of the flat-bottom distance restraints was also applied to bring the decoupled guest to a standard concentration; SOMD-4: same as SOMD-3 but with an additional correction term for electrostatic energies
Predictions were generated by either TI (labeled with TI-) or HREM/BAR (label by BAR-). Also the starting structures were obtained by quantum calculations (labeled with “ab initio”) or docking (labeled with “dock”). TI/BAR reported the lowest binding affinity from either method