How many proteins are eligible for simulation?

[steffencruz]

How can we know if a pdb is going to be a good candidate for simulation? Are the initial conditions that stop us from being able to simulate? Estimate the total number of eligible candidates.

[schampoux]

There are indeed initial conditions that stop us from being able to simulate with the intent of getting good experimental results.

• Is the force field you are using suitable for the nature of the system you are studying?
• Is the Environment in your simulation box close to the environment you would find that protein in naturally?

Not all PDB files can be simulated using GROMACS to obtain good results. The following are concerns that contribute to the suitability of a PDB file for simulation:

• Quality of the PDB structure: High-resolution structures with well-defined atomic coordinates and minimal steric clashes.
  • Steric clashes arise due to the unnatural overlap of any two non-bonding atoms in a protein structure
• Biological Relevance: Proteins with known biological functions, stable structures under physiological conditions, and relevance to specific biological processes are typically better candidates.
• Force Field Compatibility
• System Complexity: Presence of ligands, cofactors, ions, solvent molecules can all influence simulation outcomes.

[schampoux]

It may be of interest to restrict ourselves to a specific type of molecule, therefore limiting force field selection. With this approach, we will be limiting guess work for selecting hyperparameters. Rodrigo will be providing us with subsets of PDB files that have similar configurations.

[schampoux]

If the goal is energy minimization in the simulations and we wish to switch out PDB files with minimal adjustments to hyperparameters, structural similarity is the most relevant criterion.

1. Consistent Force Fields: likely to be suitable for one another because they will have similar 3d conformations and therefore similar physical and chemical environments for the atoms involved. The following parameters would not need significant adjustments: van der Waals forces, electrostatic interactions, and bond energies.
2. Minimal Setup Changes: Allows us to maintain similar simulation conditions (e.g. solvation parameters, ionic strength, temperature)
3. Relevance to Functional Insights.

• Look for proteins with low RMSD values and high TM-scores relative to your target. These indicate close structural alignment.
• Utilize the RCSB PDB's tools for structural similarity search, which allows you to put your protein's PDB id and find others with similar 3d structures.