Prediction method for signal peptide and cleavage site thereof on the basis of layered mixture model

Abstract

The invention discloses a prediction method for signal peptide and a cleavage site thereof on the basis of a layered mixture model. The prediction method comprises the following steps that: firstly, in a first layer, applying an SVM (Support Vector Machine) classifier based on amino acid residue features to identify whether a protein sequence contains N-end hydrophobic fragments or not; then, in a second layer, applying a Naive Bayes and SVM classifier based on amino acid residue features and functional structural domain features to identify whether the hydrophobic fragments are the signal peptide or N-end transmembrane helixes or not; and finally, in a third layer, according to a statistical learning rule, screening candidate cleavage sites, calculating a statistical credit score, then, calculating the similarity score of a signal peptide sequence through a Needleman-Wunsch sequence comparison algorithm, and determining a predicted signal peptide cleavage site for the statistical credit score and a sequence similarity score integral.

Description

technical field [0001] The present invention relates to a method for predicting a signal peptide and its cleavage site based on a layered mixed model, which uses known protein sequences to predict whether the protein contains an N-terminal signal peptide, and predicts its cleavage site, especially a An algorithm that fuses amino acid residues and functional domains, combines statistical confidence scores and sequence similarity scores, and predicts signal peptides and their cleavage sites hierarchically from top to bottom. Background technique [0002] In 1979, G.Blobel and D.Sabatini proposed the Signal hypothesis for the first time based on experimental observations. G.Blobel and D.Sabatini believe that there is an amino acid fragment at the N-terminus of the secreted protein sequence that acts as a signal guide, which can guide the transfer of proteins between membranes and transport proteins to their destinations. They called this segment of amino acid that guides the s...

AI Technical Summary

Problems solved by technology

Therefore, it is difficult for the classifier to correctly identify the signal peptide and transmembrane helix based on the amino acid residue characteristics, so that the false positives of the signal peptide prediction are too high

Images