Figures (3)  Tables (1)
    • Figure 1. 

      FTGD platform build flowchart. To develop FTGD, we first collected plant flowering gene datasets from two databases. Second, we extracted features, including physicochemical properties, sequence composition, and sequence order features. Third, we performed feature selection through a combination of features and dimensionality reduction. Fourth, we built seven machine learning models, consisting of three single-feature models and four combination feature models. Fifth, we conducted experimental validation through enrichment analysis and literature review. Finally, we established the FTGD database and provided online prediction capabilities.

    • Figure 2. 

      The top 15 GO enrichment charts for genes related to flowering-time in Brassica rapa.

    • Figure 3. 

      FTGD website. An overview of the FTGD database, highlighting its key interfaces and internal features, which encompass Home, Species, Download, FTAGs_Anno, Userguide, Submit, and Links interfaces.

    • MethodsNumber of
      feature
      F1-scoreACCAUC
      SVM-ACC270.7690.8110.849
      SVM-Kmer4000.8720.8900.929
      SVM-PC-PseAAC220.7660.8100.915
      SVM-Kmer-ACC4270.9190.9260.898
      SVM-Kmer-PC-PseAAC4220.9340.9390.943
      SVM-ACC-PC-PseAAC490.7920.8290.896
      SVM-ACC-Kmer-PC-PseAAC4490.8870.9010.909

      Table 1. 

      The prediction performance of SVM model.