Chloroplast structure, codon usage bias, and machine learning-based molecular identification using DNA barcoding of...

To investigate the complete chloroplast genome sequences and codon usage bias of "Shan Dou Gen" (Sophorae Tonkinensis Radix et Rhizoma) and its six easily confused species, analyze their evolutionary patterns, and evaluate the identification efficiency of four DNA barcodes combined with two machine learning methods for these seven plant species. Chloroplast gene structures of the seven species were aligned to construct phylogenetic trees. Codon usage bias was analyzed using CodonW and CUSP. Genetic distances were calculated based on the Kimura-2-Parameter model to assess the barcoding gap and reconstruct phylogenetic trees.Species discrimination capabilities of four DNA barcodes (ITS2, matK, psbA-trnH, and rbcL) were compared. Species identification was performed using BLOG and WEKA machine learning algorithms. Single-nucleotide SSRs predominated in chloroplast genomes, primarily composed of A/T bases. Complete species differentiation was achieved using cpDNA. Natural selection was the primary factor influencing codon usage bias, followed by mutation pressure. Among synonymous codons, A > T and G > C in base composition, with optimal codons ending in A/U at the third position across all seven species. All four DNA barcodes successfully discriminated Shandougen from its confusable species. The BLOG algorithm achieved >85 % identification accuracy, outperforming WEKA. This research provides a theoretical foundation for ensuring clinical medication safety, elucidating plant phylogeny, facilitating species identification, and guiding resource conservation and utilization of Shandougen and its analogues.