[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [IMP-dev] Curious Error and PDBs with more than 1 model

On Nov 1, 2009, at 11:03 PM, Javier Ángel Velázquez Muriel wrote:

Hi. I'm testing IMP with a PDB file that has 3 models inside, and is quite big: ~900 kDa, 60 chains. I was expecting problems with reading more than one model, but I'm also getting this:

ERROR: Can only allocate 150000 particles. Yell at Daniel.
As it says, there is currently a limit on the total number of particles. I can increase it if needed as, apparently, is needed.

1) Can IMP read multi-model PDBs now ? The code seems that it can.
It has a flag to read one or all models, but I'm not sure that the reading all option is very useful.

2) How are the models treated? One IMP model per PDB model ? I would like to have all the PDB models into the IMP model in this case.
Everything always goes into the same model (since the read_pdb function only takes one model as an argument). It currently just produces a atom.Hierachy node with all chains and molecules from all models as children, making sorting things out a bit annoying (as was discussed in the meeting on pdb reading).

3) How does IMP identify chains internally? How can I access chain 54, for example, given that it's name is repeated in the PDB file? (The PDB has 3 models, 20 chains each, repeated name).
They will just be repeated in the model. Not necessarily very useful.

It is probably worth producing a separate "read_all_models_from_pdb" which produces a more structured output (say a vector of Hierarchies, one per model). That would be quite easy. If we do that, we should remove the read all models flag from the existing read_pdb.