|
In our major effort this year, Jeff Roach has developed an important new method for multiple protein structure alignment and comparison based on a geometrical algorithm for reducing three dimensional structures to a one dimensional string of characters that encode structures such that the alignment tools developed for protein and nucleic acid sequences, including dynamic programming, can be used directly. Our initial applications of this method have shown it to be superior to other alignment methods based on r.m.s.ds either for coordinate superposition or for distance matrix methods. It is roughly twenty times faster, and it enables the identification of structural similarity in regions that cannot be superimposed without segmental reduction and evaluation of individual component rotation and translation parameters.
This development has involved the important contributions of Shantanu Sharma, from the ITT in Kanpur, India. Shantanu visited twice to Chapel Hill, during which time he worked on the dynamic programming application. A manuscript is nearly ready for submission and we have data for which to write a second manuscript.
We believe that this work offers a significant improvement in both the speed and in the quantitative analysis of protein structural similarity. It should permit evolutionary analysis of deeper relationships between more distantly related proteins than the many currently available methods for structural alignment and evaluation. We have already had positive readings from the Helen Berman at the PDB about the general applicability of this new program, TETRADA (TETRAhedral Decomposition Analysis). An additional objective is to determine if the quantitative scoring of alignments can improve the scoring of competitions like the CASP competition in which different competing predictions of three-dimensional structure are scored currently by a complicated scheme involving numbers of amino acids aligned together with the resulting r.m.s.d.
Finally, Jeff has been working with Ricky Armwood, a UNC undergraduate mathematician in an extension of the superposition algorithm we use to visualize the structural alignments. This work involves algorithmic development, and is potentially publishable.