Bob Hanson wrote: > Jmol users, > > Having properly implemented CONECT in PDB files, it occurs to me that > all along Jmol has not really properly handled these. This most recent > point was that a single CONECT block must apply to all models in an NMR > model set. And that is fixed. But there is an underlying issue here. > > When you load a PDB file into Jmol, certainly for most biomolecules the > CONECT block is completely ignored, and Jmol goes and creates its own > bonds instead. All CONECT information is ignored. This feature is turned > off using "set autobond FALSE" > > In reading the material Rolf sent, though, it seems clear to me that in > the case of PDB files, the CONECT information is principally for the HET > records. It's assumed that the viewer will properly bond all ATOM records. > > Q: Is that a correct reading? >
Yes, I think so. > So if you do use "set autobond FALSE" you get something that really > isn't usable. In fixing the business with the multimodel CONECT, I had > to implement bond caching -- Jmol acquires the list of CONECT > information, then applies it to all models later. This is quite > different from what Jmol was doing before, and it just might allow for > another possibility. > > Q: Would it be better if autobond respected the CONECT records of the > originating file? > > That is, should autobond not start from scratch but, in the case of PDB > files, first apply those CONECT records, THEN do the rest of the bonding? > I always thought this was the default behaviour for loading PDB files. If it is currently really not done this way I don't understand the explanation for the cross-model bonding problem any more: The CONECT record information caused it because atoms were missing in the last model so they were taken from an earlier model. But also a lot of other bonds not described in the CONECT records were displayed. And I never turned off autobond. > Q: Is there EVER a time you want to completely ignore CONECT records and > start all over? Would then somehow NOT be correct in some way? > > The penalty of implementing this is just the memory required to save the > CONECT information rather than tossing it out after the file is read. > > Advise, please. > I think it should be possible. It would at least allow to compare the autobonding result with the supplied bonds. And since there usually are not many CONECT records (on average about 60 connected atoms per entry) I wouldn't expect memory to be an issue here. Regards, Rolf ------------------------------------------------------------------------- SF.Net email is sponsored by: The Future of Linux Business White Paper from Novell. From the desktop to the data center, Linux is going mainstream. Let it simplify your IT future. http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4 _______________________________________________ Jmol-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jmol-users
