Global maximum-parsimony based ancestral reconstruction with non-universal genes

Ancestral reconstruction can be useful in many facets of genetic analysis. It allows for a look into the (potential) past, and can provide a clearer idea of species divergance. Unfortunately, ancestral construction is not as easy with swiftly changing genomes as input. Confounding factors such as horizontal gene transfer and gene loss limit available information under current heuristics that mitigate the computational difficulty of finding an exact solution. Here, we address a specific limitation of these heursitics, namely the restriction that genes be present in all leaf genomes. We show that we can greatly improve the number of genes placed in an ancestral genome by accepting genes not universal to the whole data set but universal to a given subtree. We implement this approach by modifying the popular ancestral reconstruction program PATHGROUPS and test it on fruit fly and mosquito data with multiple available genomes.