Automatic construction method of non-model species metabolism model based on protein language model

By combining blast sequence alignment and pre-training deep learning model, all proteins in cells are homologously aligned, and a variety of protein function prediction tools are integrated to annotate non-homologous genes, which solves the problem of insufficient protein homology relationship recognition ability in the prior art, and significantly improves the quality and prediction accuracy of the metabolic network model.

CN119741975AActive Publication Date: 2025-04-01SHANGHAI JIAOTONG UNIV
View PDF 6 Cites 1 Cited by

Patent Information

Application Number
CN202411938706.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-26
Publication Date
2025-04-01
Estimated Expiration
2044-12-26

AI Technical Summary

Technical Problem

The prior art cannot effectively identify protein homology relationships caused by structural similarity, resulting in poor quality of cell metabolism network models, low prediction accuracy and high cost of building models.

Method used

A combination of blast sequence alignment and pre-trained deep learning model was used to homologously align all proteins in cells, and a variety of protein function prediction tools were integrated to annotate non-homologous genes, thereby building a high-precision genome-scale metabolic network model.

Benefits of technology

It significantly improves the accuracy of gene homology relationships, improves the speed, quality of non-modal biological metabolic models, and predicts the final model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119741975A_ABST
    Figure CN119741975A_ABST
Patent Text Reader

Abstract

A non-model species metabolism model automatic construction method based on a protein language model is characterized in that a gene homologous relationship inference method based on a pre-trained large model is combined with a protein sequence comparison method to construct gene homologous relationships of different species; gEMs of a model organism are used as a template, GEMs of a non-model organism are constructed based on a gene homologous relationship and a non-homologous gene annotation result, and metabolism model structure optimization is realized through gap filling. According to the method, the accuracy of the gene homologous relationship can be remarkably improved, and the construction speed and quality of the non-model organism GEMs and the prediction performance of a final model are improved.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Automatic genomic metabolic network model modifying method

    CN104699997A

  • Method for constructing, optimizing and visualizing genome metabolism model based on high-throughput sequencing technology

    CN113035269A

  • Immunogen prediction system and method based on protein three-dimensional structure and graph neural network

    CN116206676A

  • Metabolic model de novo construction method based on protein 3D structure information comparison

    CN118230821A

  • Protein sequence analysis method, device, equipment, medium and product

    CN118571322A

Cited By

  • Genome scale metabolic network model construction method

    CN121963837A