Accepted_test

Our work is devoted to creating a gene function annotation method that works with high accuracy regardless of the age of genes. We proposed an algorithm for gene function annotation based on the search for similar sequences using the k nearest homologs method, analysis of orthogroups in the OrthoDB database, and filtering of GO terms based on sequence similarity using logistic regression. The methods were tested on sequences of protein-coding genes from a number of plant genomes, including Arabidopsis thaliana et al. Estimation of gene ages was used using the Orthoscape program. We also proposed to use a measure of prediction efficiency (PE) to evaluate the accuracy of gene function prediction methods, which avoids the influence of using different annotations when comparing different prediction methods.