Identifying biassed blocks

at the Fred Hutchinson Cancer Research Center BLOCKS WWW server.


Compositionally biassed blocks are made up of sequence segments rich in a few amino acids or short repeats. Biassed blocks are not neccesarily misaligned. Some conserved protein regions are rich in charged residues, like basic-zipper transcription factors, or in hydrophobic amino acids, like transmembranal regions. However, blocks biassed toward the same amino acid(s) can give a significant LAMA (block to block alignment) score, even without being biologically related.

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.

Paste your block(s) below and . The blocks must be in the BLOCK format. You can reformat multiple sequence alignments to that format here)


LAMA Searcher Help

[Blocks home] [Block Searcher] [Block Maker] [Get Blocks] [Blocks Formatter] [LAMA Searcher]


Contact us

Page last modified June 2000