FIG 1.

The effect of subalignment length on model performance. The y axis depicts model performance for subalignment-based models as area under the receiver operating characteristic curve (AUC) values, and the x axis depicts subalignment nucleotide length (in kilobases). Error bars represent the standard deviation of multiple random samples for each length. The number of random samples for each subalignment is shown in Table S1P in the supplemental material. A separate set of 5-fold cross-validated models was computed for each subalignment and antibiotic.