izhdb-FreeEnergyProtocol.txt

Name

multi site lambda dynamics

Software

CHARMM Developmental Version 41a2

Parameter

CGenFF force field for ligands (partitioned via MATCH)
CHARMM 36 protein field field
TIP3P explicit solvent
Assumed pH 7.0,ionic concentration 100mM (protein simulation only)
Typical simulation time per window: 20 ns
Langevin dynamics with the Leapfrog Verlet algorithm
Nonbonded cutoffs of 12 A

Method

The CHARMM 36 protein force field & CGenFF ligand force field were used with the
TIP3P water model to simulate all ligands bound and unbound to FXR. Solutes were solvated
in cubic boxes with ca. 2100 and 10000 water molecules for solvent and protein simulations,
respectively. All molecules within each Free Energy set were split between three to four
unique multi site lambda dynamics simulations (MSLD) to calculate all relative free energies
of binding. Five independent duplicate simulations were run for each, from which standard
deviations were calculated. The adaptive landscape flattening algorithm for MSLD was employed
to optimize biasing potentials prior to simulation. Simulations were performed in the NVT
ensemble at a temperature of 25 C. All simulations were performed with the CHARMM molecular
simulation software package.

izhdb-PosePredictionProtocol.txt

Name

Flexible CDOCKER (CHARMM DOCKER)

Software

Flexible CDOCKER

System Preparation Parameters

CHARMM36 force field for proteins and CHARMM General Force Field (CGenff 3.0.1) for ligands. Assuming pH 7 to assign the ligand's protanation state.

System Preparation Method

The force field for the ligands was generated using the program MATCH based on CGenff force field.

Pose Prediction Parameters

In simulated annealing, the temperature is first increased from 300K to 1000K in 6000 steps with a step size of 1.5fs. Then the temperature is decreased from 1000K to 50K in 14000 steps with the same step size. van de Waals energy switch function and electrostatic switch function are used from 12 to 14 angstrom. A distance dependent dielectric constant is used with epsilon of 3.

Pose Prediction Method

1. Inital pose preparation.
For the ligands with ID of 1 to 36, the crystal structures of protein-ligand complex are used as the starting conformations. For the other ligands, the starting conformation is obtained by docking the ligand ontfo the crystal structure of proteins which bind with similar ligands out of the first 36 ligands.
2. Flexible CDOCKER.
The starting conformation of the protein-ligand complex is first minimized. Then 10 runs of simulated annealing is used to search for the lowe energy conformations. During the simulate annealing, the protein side chains which have heavy atoms within 5 angstrom of the ligand is flexible and all other parts of protein is fixed. The end conformations from simulated annealing are used for scoring.