A Boruta-SMOTE Integrated Approach for Rapid Donkey Breed Classification Using SNP Data: Addressing High-Dimensionality and Small Sample Challenges.

With the increasing complexity of genomic data, traditional classification methods face dual challenges of the "curse of dimensionality" and class imbalance when processing multiple single nucleotide polymorphism (SNP) markers. To address these challenges, this study proposes an innovative approach integrating the Boruta dimensionality reduction algorithm with the Synthetic Minority Over-sampling Technique (SMOTE). The methodology involves two key steps: Feature optimization using the Boruta algorithm to identify the most representative genetic markers, thereby significantly reducing the complexity of high-dimensional data. Application of SMOTE technology to generate synthetic samples, balancing minority class distributions and alleviating data imbalance issues. Experimental results demonstrate that the proposed method outperforms traditional classifiers (Random Forest [RF], K-Nearest Neighbors [KNN], Extreme Gradient Boosting [XGBoost] and Convolutional Neural Network [CNN]) without Boruta-SMOTE integration across multiple metrics including accuracy, precision, recall, and F1-score. This study provides new insights for the conservation of donkey genetic resources, breed improvement, and commercial applications, while offering an effective solution for genomic data classification challenges.