Optimizing evidence theory based K nearest-neighbor alpha-helix prediction method

Abstract

The invention discloses an optimizing evidence theory based K nearest-neighbor alpha-helix prediction method and relates to correlation techniques of mode recognition algorithms and computational biology. By means of the optimizing evidence theory based K nearest-neighbor alpha-helix prediction method, a membrane protein alpha-helix structure is accurately predicted when a protein sample with a high-resolution known structure is lacked. Multiple sliding window extraction feature vectors are fused to perform optimization by adopting a computational biology method including protein multiple sequence alignment and an OETPKNN algorithm, noise is smoothed by means of a median filtering method, then a prediction result is divided by means of a dynamic threshold method, and finally the membrane protein alpha-helix structure is obtained. By means of the optimizing evidence theory based K nearest-neighbor alpha-helix prediction method, the alpha-helix prediction accuracy is improved by higher than 20%, the tail end of an alpha-helix can be predicted, and a good effect is played on irregular alpha-helixes with the prediction length smaller than 15 amino acids.

Description

technical field [0001] The invention relates to a technology in the field of bioinformatics and pattern recognition, specifically a K-nearest neighbor algorithm based on optimization evidence theory, which is suitable for predicting the protein structure of alpha helix. Background technique [0002] In recent years, with the rapid development of bioinformatics, a large amount of biological data and computer science have been combined for research and analysis, revealing more life science mechanisms endowed by biological data. Among them, proteins are an important part of biological information. The ever-expanding protein database greatly facilitates scientists' research on protein structure and function. The function of protein is often determined by its specific structure, so the study of protein structure plays a pivotal role. In the current protein database, most of the protein structures are solved by experiments, but membrane proteins have strong hydrophobicity, are no...

AI Technical Summary

Problems solved by technology

[0004] Most of the current methods can only predict amino acid sequences of regular length, such as the TOP‐PRED method (Claros MG, von Heijne TopPred II: An improved software for membrane protein structure predictions. Comput Appl Biosci 10: 685–686.) can only A protein sequence with a predicted alpha helix structure length of about 21 residues; another method based on the hidden Markov model TMHMM (Krogh A, Larsson B, von Heijne G, Sonnhammer EL (2001) Predicting tansmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 305:567–580.) cannot predict the sequence length of less than 16 residues or greater than 35 residues; SISUI method (Hirokawa T, Boon‐Chieng S, Mitaku , SOSUI: classification and secondary structure prediction system for membrane proteins. Bioinformatics) 14:378–379.) cannot predict amino acid sequences with alpha helix lengths greater than 25 residues, and these methods are currently limited to lengths ranging from 15 to 25 residues The regular alpha helix structure of the base, and it cannot be adapted to the ubiquitous alpha helix length of less than 15 or greater than 25 residues due to the complex structure and strong hydrophobicity of membrane proteins

Images