Jonathan Barnoud

The problem is not so much what DSSP would do with that file. The problem is to know if we need to do the special handling. If we have a...

This is the test indeed, but I was hoping we could get rid of the force field specificity. I'll make it work for the "universal" force field first. I have...

For the moment, only proteins are passed to DSSP. There is a `is_protein(molecule)` selector function that returns True if all the resnames are from a predefined list. At the moment,...

As you said, virtual sites do not all involve the same number of particles, and therefore cannot be stored all in one array if we store them one site per...

This would indeed be great tools to have in MDAnalysis. We would be happy to look at contributions in that direction.

To my knowledge, nobody is working on this. The issue is that MDAnalysis does not know how virtual sites are related to the rest of the atoms.

The latest solution we discussed was to encode the virtual sites as a list of duplets `(virtual site, underlying atom)` and to use consider these the same as bonds when...

Hi Bart. What you describe is the general idea. However, I would not add the virtual sites to the bonds themselves. Indeed, while they are connections, they are not bonds...

@utkbansal The files are there already. They have been built with the software that usually generate them (simulation engines mainly). If I well remember, the test classes have a bunch...

Among the file formats, find one that you understand (the file format, not the code at first). Then see if you understand the code that you are going to test:...