Five randomly chosen example segmentations on simulations from our simple bidirectional introgression scenario with relatively strong positive selection acting on each introgressed nucleotide (s = 1 × 10⁻⁵, or ∼0.095 for a 10 kb introgressed segment).

Each example shows the input alignments for the two populations (labeled “pop 1” and “pop 2” respectively), the true introgressed alleles for these two populations (labeled “pop 1 (y)” and “pop 2 (y)” respectively), the introgressed alleles inferred by IntroUNET (“pop 1 (pred)” and “pop 2 (pred)”), and IntroUNET’s inferred introgression probabilities (labelled “prob”). Alignments and introgressed histories, true and predicted, are shown in the same format as in Fig 1.

(PDF)