SplitAligner: A Gene-Species Tree Reconciliation Framework Using Split-Based Branch Mapping
Wu, J.
Show abstract
Phylogenomic analyses are increasingly focused on branch-specific questions within a fixed species tree. However, two pervasive challenges in real datasets--missing taxa and gene-tree/species-tree discordance--complicate the comparability of branches across loci. Here, we introduce SplitAligner, a split-based framework that defines branch identity on a fixed species-tree backbone and evaluates it gene by gene under varying taxon coverage. For each species-tree branch, SplitAligner projects its split onto the gene-specific taxon set to determine whether the branch is evaluable or structurally missing due to a degenerate projected split. Under fixed-topology gene trees, this projection reveals branch fusion, where multiple species-tree branches collapse into an indistinguishable fusion group on the observed taxa. SplitAligner reports such cases as composite fused-branch identities and aggregates branch lengths accordingly. Under free-topology gene trees, SplitAligner further identifies topology-induced missingness (NA_topo), where a branch is decisive under the projected-split criterion but its projected split is absent from the gene tree, separating discordance-driven absence from coverage-driven missingness. These operations produce standardized gene x branch tables and a branch-wise concordance score (Support), defined as the fraction of decisive genes whose free-topology trees recover each projected split. Applying SplitAligner to 2,275 single-copy genes from a dataset of 302 mammals reveals heterogeneous concordance across the mammalian phylogeny and highlights internodes with elevated discordance-associated missingness. The resulting branch coordinate system provides a general framework for branch-based estimates of evolutionary rates, selective constraints, and other branch-wise summaries across thousands of loci/genes.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.