1zna: EPPIC 3.0-alpha does not stop with non-standard SGs and settings

[gcapitani]

For entry 1zna (a Z-DNA tetramer) in non-standard SG B 2 21 2 EPPIC 3.0-alpha (unlike 2.1.2) does not issue the "non-standard space group error" and completes the run just calculating the AU interface. This version does not yet check for non-standard SGs and settings, which is needed to keep out several hundreds of affected entries in PDB-wide precomputations.

[sbliven]

This can be solved very similarly to #133

[josemduarte]

That's a good idea, I think we could simply add a new "nonStandard" flag to the PdbInfo table to be able to display this and #133 consistently. The flag would then be populated from BioJava. I can do that part.

[sbliven]

Ok, so we will implement this as follows:

1. Add nonStandard boolean column to PdbInfo. For testing purposes, this could be filled with random data or whatever
2. In the GUI, add a warning to the structure warnings list if this boolean is true. Warning should say "Non-standard space group. Only contacts in the asymmetric unit will be analyzed."
3. During database upload, the space group should be checked against the list of valid space groups. If it is not on the list, set nonStandard=true

[josemduarte]

So I'm working already in the server side of this. Pull request biojava/biojava#604 is a step towards this.

I think we should be able to capture 2 sub-cases here:

• non-standard spage groups, e.g. B 2 21 2 in 1zna
• non-standard coordinate frame convention, e.g. 4hhb and all those discussed in Be more forgiving of non-standard crystal frames owl#4. This is detectable by comparing the scale matrix in the file to our calculated scale matrix.

I will add the nonStandard column to PdbInfo and set it to true in either of the 2 cases above. So I will take care of points 1 and 3 in @sbliven's list. Point 2 I will leave to @altheaparker. Only a small addition to the warning message: "Non-standard space group or setting. Only contacts in the asymmetric unit will be analyzed."

[josemduarte]

I've now done 1 and 3.

I've actually added 2 fields in PdbInfo instead of just one:

• nonStandardSg : warning for this should be "Non-standard space group. Only contacts in the asymmetric unit will be analyzed."
• nonStandardCoordFrameConvention : warning for this should be "Non-standard coordinate frame convention. Only contacts in the asymmetric unit will be analyzed."

I've put the commit in branch issue37. It depends on this PR in biojava: biojava/biojava#605 (which also needs to be cherry-picked into biojava's bugfixes-4.2)

[josemduarte]

I've now merged the branch with these changes. Note that it depends on biojava-4.2.5-SNAPSHOT (available from the bugfixes-4.2 branch).

[sbliven]

Is it possible to convert the dev db to have these columns, or do we need to wait for the next repopulation before testing this?

[josemduarte]

Running any of our scripts (with setting hibernate.hbm2ddl.auto=update, which is our default) against the database will update the schema, setting the fields to false. One can then set some to true with a manual update command.

An alternative is doing a full run for 4hhb, produce the webui.dat file and then upload it with UploadToDb (with force replacement option).

[josemduarte]

I've just run the UploadToDb script in dev db, so the schema is updated.

[josemduarte]

Looks like this works now, thanks @altheaparker! Should we close?

[sbliven]

Looks good. The database needs to be repopulated, but manually setting values (3hbx and 1zna) shows that it works.