minor issues about hdf5 files

[LilySnow]

1. Maybe we should not call the pdb file of a model as "native" in the BM4 hdf5 files (e.g., 1E6E.hdf5) and call it "pdb" instead:

In [8]: list(f['1E6E_9w']) Out[8]: ['complex', 'features', 'features_raw', 'grid_points', 'mapped_features', 'native', 'targets']

2. Shall we also put haddock score in the BM4 hdf5 files for easy comparison with haddock scoring function?

3. In the final output data.hdf5 file, shall we also store the model IDs (currently it only contains target DockQ and predicted dockQ). The current version of data.hdf5 is not convenient for the comparison with other methods.

[NicoRenaud]

'complex' is the pdb of the conformation (e.g. 1E6E_9w.pdb) and 'native' is the corresponding native conformation (e.g 1E6E.pdb). That being said we can rename all of that that very easily.

[LilySnow]

Inside 1E6E.hdf5, we already have a folder for the native, as below, right? Should we then remove the "native" entry for all models inside 1E6E.hdf5, since they seem to be redundant?

In [6]: list(f['1E6E']) Out[6]: ['complex', 'features', 'features_raw', 'grid_points', 'mapped_features', 'native', 'targets']

[NicoRenaud]

we can 'clean' the hdf5 file and remove all entries that are not needed But this is at the cost of possibly not being able to add new data to it

[LilySnow]

Sorry, I do not understand. Why we have to have a cope of the native pdb file in each of the model file?

[NicoRenaud]

It's needed to compute i-rmsd l-rmsd and dockQ. And for convenience it's stored there as well but can be removed if needed

[LilySnow]

But the pdb file of the native is already in 1E6E.hdf5 as a separate entry, right?