Conversation
| raise NotImplementedError() | ||
|
|
||
| @property | ||
| def filtered_correlation_matrix(self): |
There was a problem hiding this comment.
I don't see the reason to have this function - couldn't this step just be performed in self.run?
also, it would be nice if we could have a low and high threshold, and keep values that are > high and < low. This is because we might want correlations that are larger in absolute value than a certain threshold (and keep the highly negative corrs)
| def __init__(self, ref: str, traj: str, atoms: str, backbone: bool) -> None: | ||
| self.ref = ref | ||
| self.traj = traj | ||
| self.selected_atoms = "backbone" if backbone else f"name {atoms.replace(',', ' ')}" |
There was a problem hiding this comment.
usually these measures are calculated on just CAs rather than backbone.
Even then, it would be good to have e.g. a default selection option (all CA atoms in protein) or the possibility to give a custom MDAnalysis selection string, akin to what we do in groups for hydrogen bonds
| (".trj, .pdb, .xtc, .dcd"), file_name).validate_file_extension(), | ||
| required=True) | ||
| common_args.add_argument("--threshold", type=float, default=0, help="Threshold for the correlation analysis") | ||
|
|
There was a problem hiding this comment.
I think a step is missing here - before calculating DCCM we should best superimpose each trajectory frame to one reference frame (usually the topology) to eliminate roto-translations as much as possible. This fit procedure should also allow to define custom groups to select atoms/residues to fit on (default could be still all CA)
2 participants
No description provided.