Machine Learning-Based Prediction of Orphan Genes and Analysis of Different Hybrid Features of Monocot and Eudicot Plants

Orphan genes (OGs) may evolve from noncoding sequences or be derived older coding material. Some shares of OGs are present in all sequenced genomes, participating the biochemical and physiological pathways many species, while them associated with response to environmental stresses species-specific traits regulatory patterns. However, identifying is a laborious time-consuming task. This paper presents an automated predictor, XGBoost-A2OGs (identification for angiosperm based on XGBoost), used identify seven species hybrid features XGBoost. The precision accuracy proposed model fivefold cross-validation independent testing reached 0.90 0.91, respectively, outperforming other classifiers cross-species validation via models, namely, Random Forest, AdaBoost, GBDT, SVM. Furthermore, by analyzing subdividing into five sets, it was proven that different feature sets influenced prediction performance involving eudicot monocot groups. Finally, small-scale empirical datasets each separately optimal revealed performed better groups than