Appropriate threshold setting and multiple methods combination may improve reproducibility of gene ontology enrichment analysis

Transcriptome analyses are widely used for biological research. Gene ontology (GO) enrichment analysis is often used for effectively analyzing large data matrices. However, various software and methods can be used for performing GO enrichment analyses, which might lead to different conclusions. To date, there is no agreement whatsoever in the scientific research community about standards and processes for analysis; moreover, the description about such analyses in previous research is often brief, causing difficulties in both research reproducibility and manuscript review. Herein, we introduce the advantages of different tools and methods, while the limitations or problems are stated. We found that the essential key to improving the reproducibility and accuracy of GO enrichment analyses was to set appropriate thresholds in data processing and might combine different GO methods to avoid the limitations of each one. Finally, research associations might need to consider draft-standardized generic transcriptomic analysis standards to promote advances in biological research. • Different GO enrichment analysis methods may lead to different conclusions. • No standards for GO enrichment analysis caused difficulties in reproducibility. • Setting appropriate thresholds and combining analysis methods are essential.