Fig. 1

Sensitive pan-genome search for MHC class I genes. a The canonical domain structure of MHC class I proteins and (b) genes. c The location in the opossum genome and score of matches to profile hidden Markov models representing the antigen-presenting domain (split into α₁ and α₂ regions), C-type immunoglobulin domain and C-terminal domain. d Example of a high-scoring run of α_1, α₂, Ig and C-terminal domains in the opossum genome. e Finite state automata of the alignment algorithm to search for runs of α_1, α₂, Ig and C-terminal domains, taking domain score and distance between domains into account. The nodes (circles) show match states. Symbols on edges show scores/penalties: +m is the match score, which is based on the HMM match score; -γ is a distance-dependent affine gap penalty, which models introns and allows the alignment to skip over matches that interrupt a run of domains; -ψ is a constant penalty for dropping the C-terminal domain