Reference mapping and whole-genome alignment yield comparable phylogenomic inferences

The reference genome selected for genome mapping can inadvertently introduce bias when analyzing sequencing data against a non-kin species genome. However, the extent to which this bias influences species-level phylogenomic inferences remains uncertain. To investigate this, researchers conducted a study across Ursidae, a family of bear species, using nine species spanning around 20 million years of divergence. This group included the extinct giant short-faced bear (Arctodus simus).

The team mapped the same sequencing datasets to four reference genomes with varying degrees of evolutionary distance and compared the results with two assembly-based whole-genome alignment approaches. To streamline the analysis, they developed PhyloSlide, a sliding-window phylogenomics pipeline. Applying this method across all treatments, they discovered that the choice of reference genome significantly impacted sequence recovery.

Missing data tended to increase with evolutionary distance, suggesting that older, more divergent species were more affected.

Despite this reference-dependent sequence loss, all six treatments successfully recovered an identical species-tree topology. They also found highly similar gene and site concordance factors, indicating that the phylogenomic inferences were remarkably consistent across the different methods. Relative divergence-time estimates were also largely consistent, further supporting the robustness of the findings.

In conclusion, the study highlights that genome mapping from multiple species to a shared reference can produce phylogenomic inferences comparable to whole-genome alignment. This finding holds even when substantial reference-dependent sequence loss occurs over millions of years of evolutionary divergence, suggesting that the choice of reference genome may not significantly impact species-level phylogenetic analyses.