Protein interaction prediction method

Abstract

The invention discloses a protein interaction prediction method based on a sampling strategy of non-interaction protein pairs fused with biological semantics. Protein pairs in different molecular functions, biological processes and cell components are sampled and combined based on GO term semantic similarity to obtain an NIPs subset. With a negative set sampling strategy, a non-protein interactiondata set with higher quality and low selection deviation is obtained, so that a protein interaction prediction model with better robustness and better prediction performance is obtained through training.

Description

technical field [0001] The invention relates to the technical field of biological information, in particular to a protein interaction prediction method based on a sampling strategy of non-interacting protein pairs fused with biological semantics. Background technique [0002] Protein-protein interactions (PPIs) play an important role in the structure and function of cells, and the study and reconstruction of PPIs network will not only help to understand cellular processes and disease pathogenesis, but also help to develop therapeutic drugs . Existing experimental methods for PPIs have labor-intensive and time-consuming limitations, leading to the need for computational prediction of protein interactions. Although some advanced PPIs calculation and prediction models have been proposed, most of the calculation models require both positive and negative samples for model training, which requires high-quality PPIs and NIPs (non-interacting proteins) data. Currently, PPIs valida...

AI Technical Summary

Problems solved by technology

This sampling method reduces the false negative rate, and the obtained negative data is more reliable, but it cannot cover NIPs in the same subcellular location, resulting in selection bias in the model prediction, and it is difficult for the generated model to obtain better results in real work scenarios

Images