Something that has not been mentioned is getting the sequence out of the density.  If you can trace and assign a stretch of 6 or so residues it tends to be rather unique in the PDB.  20^6 is a big number.  Several people whose names I will not mention have "discovered" the protein they actually crystallized this way.

-James Holton
MAD Scientist

On 11/16/2015 8:11 AM, Graeme Winter wrote:

Dear All,

Thank you to everyone who replied so quickly - seems it was not such an odd request after all!

First thing past the post was PDBeFold - which seems to be what msd-fold evolved into, that hit the spot in several seconds after I uploaded the shelxe chain trace. Interestingly it looks like it found "the right thing" however the structure is deposited in P21 (completely different cell) where the structure solution was in P42212, so turns out computers and databases *work*

Awesome :o)

Cheerio Graeme

-----Original Message-----
From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Graeme Winter
Sent: 16 November 2015 15:36
To: ccp4bb
Subject: [ccp4bb] Odd monday afternoon question: can I search the PDB for a structure

Hi Folks

An unusual question for you: if I have an electron density map, is there a way to search the PDB for structures which match it? i.e. they may be in a different space group etc but I would like to see if the fold is known *with no sequence information*. Ideally searching by 3D homology with a C-alpha trace or something would be ideal.

Is there a way to do it? I appreciate that this is rather back to front with respect to the usual problem, more kin to an inverse phase problem :)

Context: I have an old data set lying around which phases nicely does not match any known unit cell in the PDB; I want to find out what it is without having to sequence by hand

best wishes Graeme