Integrating DNA barcoding and machine learning for species identification: Comparative genomics and codon usage bias...

This study integrates chloroplast genome comparison, codon usage analysis, machine learning, and DNA barcoding to elucidate the phylogeny, genetic diversity, and species identification of Gentiana Sect. Cruciata. Perform chloroplast genome analysis using IRscope (boundary analysis), MISA (SSR detection), and mVISTA (variation alignment). Based on ChiPlot, CodonW, and CUSP analysis, factors influencing codon preference and usage patterns were studied. Molecular identification based on ITS2, matK, ITS, psbA-trnH barcode with BLOG, WEKA machine learning algorithms. Chloroplast SSRs dominated by A/T repeats; non-coding regions exhibited higher variability. Codon bias driven by natural selection, with A/U preference at the third position. ITS2 showed the highest discrimination power (matK > ITS > psbA-trnH). Machine learning (J48/SMO classifiers) achieved 83.33%-100% accuracy using four barcodes. This study provides theoretical foundations for conservation, medicinal quality control, and resource authentication of the Gentiana Sect. Cruciata.