research advances
December 2009 technical highlight
Learning from failure
PSI-SGKB [doi:10.1038/th_psisgkb.2009.56]
With more than 200,000 protein targets there are bound to be a few proteins that just won't yield structures. To avoid repeating a costly failure, check our databases before you start.
SDS-polyacrylamide gel electrophoresis is used to follow the progress of protein purification. Credit: Marta Ferreira.
One thing we have learned from structural genomics is that solving protein structures is a complicated business. Despite the staggering productivity of the largest structural genomics groups, who deposit around 200 structures a year, structural genomics has not lived up to early, perhaps unrealistic, expectations.
Anecdotally, researchers point to difficulties in obtaining protein crystals as the most significant 'bottleneck' for structural biology; yet the experience of the Protein Structure Initiative (PSI) and other groups is that producing suitable proteins for structural studies is a multi-step problem, and failure can occur at any stage.
The PSI JCSG estimates that more than 60% of the cost of structure determination is due to failed experiments, making well diffracting crystals more expensive than diamonds, according to a recent paper by Wladek Minor's team, members of PSI MCSG and CSGID.
Structural genomics centers have collected a great deal of information about failed experiments, as well as successful ones, and so it is possible to correlate protein properties with success. For example, an analysis of the likelihood of crystallization reported that the most important protein factor for crystallization is the presence of well-ordered surface residues.
Where can the results of failed experiments be found? Unusually, structural genomics consortia publish both their successes and failures, as part of an agreement with the International Structural Genomics Organization (ISGO). Basic information about the project chosen is shared in the database TargetDB and these data have been used in numerous research studies. PepcDB is an enhanced version of this database and it contains information about protocols and experimental details. Both are now part of the Structural Genomics Knowledgebase.
So instead of benefiting only from our own failures we can now all benefit from each other's.
Related articles
One system to express them all
Crystals from recalcitrant proteins
Ten tips for crystallizing membrane proteins
