Proposed new modular BAC design

Our current plan is that we divide the process of running a simulation into three main steps:

1. Building (System Preparation and MD engine input generation)
2. Execution
3. Analysis

We are focusing BAC2 on the first and last of these. We however want to store (and where necessary extract) metadata to facilitate further analysis and decentralised storage.

System building has been further broken down into system preparation and the generation of run configuration scripts for different MD engines.

Initial development priorities in bold

System Preparation (Builder)

Drug Parameterization

• Initially target GAFF/AM1-BCC with ANTECHAMBER
• Start from mol2, pdb or SMILES
• Select protonation state
• Need to create hybrid topologies (TIES)

Model completion

• Model missing loops
• Apply biological unit (BIOMT)
• Create from new PDB format XML/mmCIF
• Parameterize or replace non standard residues

Simulation system preparation

• Apply user specified mutations to proteins
  • Insertions and deletions
  • Specify by target sequence (or relative to reference)
• Homology modelling
• Select residue protonation states
• Incorporate disulphides
• Ions
  • Load parameters if necessary
  • Option to build binding site using explicit bonds?
• Solvate
  • Counter ions
  • Select pH [New] ?
  • Select geometry [New] ?
• Prepare topology
  • AMBER
  • CHARMM
  • GROMOS
  • OPLS
  • Joint/multiple topologies for TI

Generate simulation configuration/run files

Vanilla MD

• NAMD
• OpenMM
• CHARMM
• AMBER
• GROMACS

Alchemical

• NAMD
• OpenMM
• CHARMM
• AMBER
• GROMACS

MIST

• GROMACS

Drug docking

Metadata/Run Tracking

• System specification (forcefield, sequence etc.)
• Extract version of simulation code from logs

Analysis

• MMPBSA
• Normal mode
• TI
• ABFE
• Structural analyses
  • User specified scripts?
  • Standard analses
• Automated equilibration analysis example code
• MSM