The accuracy of protein structure alignment servers

Abstract

Background: Protein structural alignment is one of the most fundamental and crucial areas of research in the domain of computational structural biology. Comparison of a protein structure with known structures helps to classify it as a new or belonging to a known group of proteins. This, in turn, is useful to determine the function of protein, its evolutionary relationship with other protein molecules and grasping principles underlying protein architecture and folding.

Results: A large number of protein structure alignment methods are available. Each protein structure alignment tool has its own strengths and weaknesses that need to be highlighted. We compared and presented results of six most popular and publically available servers for protein structure comparison.  These web-based servers were compared with the respect to functionality (features provided by these servers) and accuracy (how well the structural comparison is performed). The CATH was used as a reference. The results showed that overall CE was top performer. DALI and PhyreStorm showed similar results whereas PDbeFold showed the lowest performance. In case of few secondary structural elements, CE , DALI and PhyreStorm gave 100% success rate.

Conclusion: Overall none of the structural alignment servers showed 100% success rate. Studies of overall performance, effect of mainly alpha and effect of mainly beta showed consistent performance. CE, DALI, FatCat and PhyreStorm showed more than 90% success rate.