Sources, numbers, and applied filtering of training datasets.

Details for the source of each training dataset are given as reference to relevant publications. Numbers of TP and TN indicate the number of composite alignments used to train classifiers. Details on the composite alignment generation and filtering procedures for each dataset are available in the Supplementary text in S1 Appendix. The INDELible program uses a control file which we provide as S8 Dataset. All composite alignments are available at https://apollo2.chemistry.gatech.edu/TwinConsDatasets/.