Differentiation of Vector Species Using COI-Based Data: A Case Study of Simuliidae (Diptera) with Multi-Class Machine Learning

DNA sequence identification of selected vector species complexes of Simuliidae (Simulium equinum, Simulium metallicum, Simulium callidum, Simulium damnosum and Simulium ochraceum) was tested via machine learning methods. The cytochrome oxidase I gene region was used for comparisons. Sequence data obtained from the NCBI database were converted to digital format with 2-bit one-hot coding. SVM (Support Vector Machines), LDA (Linear Discriminant Analysis), Logistic Regression, and Random Forest algorithms were applied to classify the data. The 5-fold cross-validation method was used to compare the algorithm performances. The SVM model achieved the highest accuracy as a robust classifier for DNA-based species identification.