The context-dependence of amino acid properties. Intelligent Systems in Molecular Biology

One of the current]imitations f using sequence align-ments to identify proteins with similar structures is that some proteins with similar structures do not have significa~ut sequence similarity by identity. Oneway to address this "hidden-homology " problem is to match amino acids based on their chemical and physical properties. However, the amino acid properties over-lap, creating orthogonal dimensions f imilarity, he relative strengths of which are ambiguous. It has been observed that the role an amino acid plays (and hence the property hat is important) a a site in a protein depends on its secondary and tertiary environment. To approximate and take advantage of this depen-dence on context for improving the sensitivity of align-ments of proteins whose structures are unknown, we propose a surrogate definition of context based on the pattern of hydropathy in a small window of contiguous neighbors surrounding each amino a~d. We present the results of an experiment in which a search-based program iteratively tests and selects various properties in independent contexts, and incrementally increases the ability ofsequence alignments to detect relation-ships among distantly-related proteins. The method is shown to perform better than using the MDM78 substitution table for partial match scores