Additional file 2 of Machine learning-based prediction of antimicrobial resistance and identification of AMR-related SNPs in Mycobacterium tuberculosis

Supplementary Material 2: Table S1. Publicly available whole genome sequencing datasets used in this study. Table S2. The number of isolates susceptible or resistant to antibiotics and the total number of SNPs in each group used for training and testing of antibiotic-specific machine learning models. Table S3. Identification of the top three best models (based on F1-score) for each MTB-drug group. Table S4. Percentage of correctly identified resistance phenotype to the specific drug in an independent dataset of MTB isolates by 12 machine learning algorithms. Table S5. The details of potential resistance-related SNPs in each antibiotic. Table S6. Assessment of the performance of the optimal machine learning algorithms in predicting resistance to four first-line drugs in the external independent MTB isolates. Table S7. Comparative Evaluation of MTB Isolate Resistance Phenotype Prediction between Our Machine Learning Framework and GenTB Using two Independent Datasets.