PROTEIN THREADING METHODS

我们已经开发出比较蛋白质的计算机方法

We have developed computer methods to compare a protein's sequence with a library of "folds" from the structural database. The sequence is "threaded" through alternative structures, and those most compatible are identified by energy calculations, using contact potentials. Since they directly detect structural similarity, threading methods can identify very distant evolutionary relationships that may be undetectable by sequence comparison. Research this year has focused on testing of the Gibbs sampling threading method, in blind predictions and control experiments, and on algorithmic improvements to increase sensitivity. A notable success in blind prediction was the identification of human leptin as a member of the helical cytokine family, published prior experimental verification. Control experiments, using known structures, identified thresholds for successful fold recognition and accurate modeling: the similar "core" substructure must comprise 60% or more of the sequence, and must superpose to a residual of 2.5 Angstroms or less, such that contact patterns are preserved. Structural similarity can be less extensive in some cases of distant relationship, however, and several improvements to increase sensitivity have been considered. A modified contact potential adding local-structure information has been shown to produce improved threading alignments. New definitions of the "core" of database structures, according to the regions superimposable in homologs with known structures, has been show to reduce false negatives in threading predictions. Changes in the Gibbs sampling alignment algorithm itself, to detect local minima and test for convergence, have improved the accuracy of statistical significance calculations. With these improvements fold recognition may be expected to reliably detect a greater proportion of the distant evolutionary relationships, a possibility that is being tested with blind predictions for the 1996 Asilomar workshop.