Abstract: It is a standard procedure to compare new amino acid sequences to databases of proteins that have been studied already in order to find similarities in structure and function. This comparison can be sequence–sequence or sequence– structure based. In order to compare, an alignment is performed of the target protein sequence (whose structure we are searching) with a template protein (whose structure we know). For a sequence–sequence alignment, the alignment algorithm optimizes a certain scoring function that quantifies the similarities of the amino acids at individual positions. For a sequence–structure alignment, also known as threading, usually the scoring function that is optimized is designed to capture the essence of structural similarity among proteins. These scores are supposed to be comparable between different proteins, since we want to select the template which achieves the highest alignment score to the target protein as our candidate for the structural model of a protein. The involved scoring functions are inaccurate, however. Thus, it is very helpful if the method can augment the generated alignment and its score with a statistical significance value which captures the confidence that we can put into the generated alignment. While important for protein alignment and threading as stand-alone tools, significance scores are even more essential if protein alignment is used in an automated cascade of tools for protein structure prediction. In this paper we analyze the performance of several variants of the 123D protein threading method Alexandrov, Nussinov, & Zimmer and compare it to several variants of optimal sequence alignment. Where theoretically available, we analyze the statistical significance of scores. For the other methods, we propose empirical approximations to pvalues and evaluate their validity.
Publication Year: 2001
Publication Date: 2001-01-01
Language: en
Type: article
Access and Citation
Cited By Count: 1
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot