search

tubiana / TTClust

clusterize molecular dynamic trajectories (amber, gromacs, charmm, namd, pdb...)
Other
106 stars 36 forks source link

Ligand poses clustering #27

Closed Marwa-Abdelkader closed 3 years ago

Marwa-Abdelkader commented 3 years ago

If I want to cluster the ligand poses instead of the protein backbone , should I select it from -sa, -sr, or -st?

tubiana commented 3 years ago

Hi !

-sr is for calculating the RMSD matrix (used for the clustering) and -sa is the selection for the alignment of the trajectory. So this is a good question because depending the alignment of the trajectory, your will observe very different things.

  1. If you use -sa "resname LIG" -sr "resname LIG" you will aligned the trajectory on the ligand and compute the trajectory on the ligand. It means that you will cluster your trajectory based on the atomic fluctuations of the ligand.
  2. But if you use -sa "backbone" -sr "resname LIG", you keep the alignement on the protein and you will cluster on the ligand pose. This could be a bias if you just want to cluster the differents conformation of your ligand, but it will be more accurate for ligand pose. Because the ligand position will be different if your take as reference the protein.

To summary:

Best regards, Thibault.

Marwa-Abdelkader commented 3 years ago

Hi !

-sr is for calculating the RMSD matrix (used for the clustering) and -sa is the selection for the alignment of the trajectory. So this is a good question because depending the alignment of the trajectory, your will observe very different things.

  1. If you use -sa "resname LIG" -sr "resname LIG" you will aligned the trajectory on the ligand and compute the trajectory on the ligand. It means that you will cluster your trajectory based on the atomic fluctuations of the ligand.
  2. But if you use -sa "backbone" -sr "resname LIG", you keep the alignement on the protein and you will cluster on the ligand pose. This could be a bias if you just want to cluster the differents conformation of your ligand, but it will be more accurate for ligand pose. Because the ligand position will be different if your take as reference the protein.

To summary:

  • Atomic fluctuation of the ligand (local fluctuation, fragment fluctuation) -> Alignment on the ligand -sa "resname LIG" -sr "resname LIG"
  • How the ligand move in the protein -> Alignment on the protein -sa "backbone" -sr "resname LIG"

Best regards, Thibault.

Thank you so much for the clarification and for that very useful tool :)

tubiana commented 3 years ago

You're welcome :)

Feel free to open a ticket again if you have any issues.