Fig. 3

Comparison of classification performance using cross-validation of TransGeneSelector and Random Forest in germination-related task. a TransGeneSelector Performance Across Different Numbers of Generated Samples: This panel illustrates the classification performance of TransGeneSelector with varying numbers of synthetic samples. The blue line represents the cross-validation performance of the model without additional classifier filtering, and the orange line signifies the cross-validation performance with additional classifier filtering. The shaded regions around each line indicate the corresponding error bands, providing a visual representation of the uncertainty associated with each measurement. b Random Forest Classifier Performance with Varied Gene Selection: This part demonstrates the classification performance of the Random Forest classifier for different numbers of genes selected through the wrapper method. The parameter ‘n_features_to_select’ within the RandomForestClassifier module is set uniformly spaced between 1 and 500, with a specific value chosen as 200, allowing for a detailed exploration of the effect of gene selection on classifier performance