Tuesday, 15 January 2008

CCP4 2008 Study Weekend

CCP4 (Collaborative computing Project 4) is responsible for one of the main X-ray crystallography computer packages. It holds a study weekend the first weekend in January each year attended by 3-500 delegates. It is the social event of the UK protein crystallographers calendar as well as being the best methods meeting each year certainly in Europe. This year the topic was "Low Resolution Structure Determination and Validation" inspired by the retraction in Dec 2006 of five high profile structures of membrane proteins, which had been wrongly determined due to the author using an incorrect piece of software (not CCP4!) which meant that his maps were inverted ie his helices were all left handed rather than right handed. Because the resolution was low, as membrane protein crystals often are, this mistake was not immediately obvious from the maps and he forced through the refinement of right handed helices into left handed density by some dubious methods. Less was said about another structure published in Nature which essentially does not have the copies of the protein touching in the crystal as this is still under investigation by the American University from where this structure originates. Analysing these mistakes should help the community to avoid them themselves and possibly spot them as referees- all these papers had got past expert referees.
On a more positive note the first example of an ab initio (ie without a sequence homologue) prediction of a protein structure that was good enough to solve a crystal structure was presented. The main conclusion of the meeting is that the resolution of the structure is pretty much the only thing that determines the quality of the structure. Big structures at the same resolution are just as good as small structures. However big structures tend not to be at as high a resolution so that on average big structures are less well determined than small structures, but this is because they are lower resolution. Structural genomics groups are no better or worse than targetted labs at determining structures.

From a PPS point of view probably the most interesting talk was by Chris Tate from MRC in Cambridge who said that the retracted structure had to be wrong because it was incompatible with the biochemical data. Although structure is powerful, it has to be compatible with the biology. The system he worked on (EmrE) has another peculiarity in that half the protein inserts into the membrane in one direction (N terminus in) and half in the other (N terminal out) and the active molecule consists of one of each of these protein chains.

No comments: