at the Fred Hutchinson Cancer Research Center BLOCKS WWW server.
Biassed blocks have a high number of columns similar to one another. Such a block can be identified by comparing all of its columns with each other, clumping those columns sharing high pairwise scores, and counting the number of columns in the largest clump.
Two parameters are used: which pairwise column score is considered "high" and what fraction of columns in the largest clump is considered "high". The column comparison measure used is the normalized sum of products (p). p(A,B) is the sum of A(i)*B(i) for all i amino acids (20) in block columns A and B (actually PSSM columns). About 5% of the blocks in the Blocks+ (10 June 2000) database were judged biassed using the default parameters of 0.50 for the cutoff pairwise column score and 25% of columns in the largest clump of columns.
[Blocks home] [Block Searcher] [Block Maker] [Get Blocks] [Blocks Formatter] [LAMA Searcher]
Page last modified