Method for predicting potential sensitization in protein

Abstract

The invention provides a method for predicting potential sensitization in protein. The method includes: firstly, making a training positive set and a training negative set; secondly, encoding various attributes of protein to build characteristic vectors; thirdly, sorting characteristics by maximum correlation and minimum redundancy methods, and selecting optimal characteristics by progressive characteristic selection method; and fourthly, counting and analyzing the selected characteristics, and giving a report of characteristic results evidently related to the sensitization in protein. The potential sensitization of the protein can be predicted effectively, the method is more accurate than the existing computational biology prediction methods, protein characteristics related to the sensitization of the protein can be analyzed effectively, and the method is significant to prediction of allergens, and study on the sensitization of the protein.

Description

technical field [0001] The invention relates to a computational biology prediction method for protein properties, in particular to a prediction method for potential allergenicity of proteins. Background technique [0002] Allergies and other hypersensitivity reactions to food and environmental factors are a major cause of chronic disease, affecting approximately 25% of the world's population. Allergens include proteins in food, cold air, hot air, ultraviolet rays, metals, etc. Among them, allergenic proteins may cause great harm to human health. In addition, more and more genetically modified foods have entered our daily life, and the potential risk of food allergies has also increased. Therefore, it is necessary to evaluate and predict the potential allergenicity of proteins. [0003] At present, there are three main methods of allergen prediction in computational biology, one is sequence-based method, the other is motif (motif)-based method, and the third is SVM (support...

AI Technical Summary

Problems solved by technology

The sequence-based prediction method proposed by FAO / WHO (Food and Agriculture Organization of the United Nations / World Health Organization) is based on the principle of distinguishing the amino acid sequence similarity between the protein to be tested and the known allergen protein. This method can effectively The allergen protein is predicted, but its false positive rate is very high; the motif-based prediction method is to compare the protein to be tested with the allergen characteristic motifs. Compared with the sequence-based method, this method improves the specificity to a certain extent and reduces the The false positive rate, but the overall accuracy rate is only about 65%, which is far from the actual demand; the machine learning method based on SVM first establishes a data set of known allergens and non-allergens, and then calculates their amino acid composition. The amino acid composition is input as a feature into the SVM for training, and finally a training model (classifier) ​​is formed. According to the amino acid composition of the protein to be tested, the classifier judges the output result. This method greatly improves the accuracy, but it cannot know the characteristic information of the allergen

Images