Evaluating the Performance of Widely Used Phylogenetic Models for Gene Expression Evolution

Abstract

Phylogenetic comparative methods are increasingly used to test hypotheses about the evolutionary processes that drive divergence in gene expression among species. However, it is unknown whether the distributional assumptions of phylogenetic models designed for quantitative phenotypic traits are realistic for expression data and importantly, the reliability of conclusions of phylogenetic comparative studies of gene expression may depend on whether the data is well described by the chosen model. To evaluate this, we first fit several phylogenetic models of trait evolution to 8 previously published comparative expression datasets, comprising a total of 54,774 genes with 145,927 unique gene–tissue combinations. Using a previously developed approach, we then assessed how well the best model of the set described the data in an absolute (not just relative) sense. First, we find that Ornstein–Uhlenbeck models, in which expression values are constrained around an optimum, were the preferred models for 66% of gene–tissue combinations. Second, we find that for 61% of gene–tissue combinations, the best-fit model of the set was found to perform well; the rest were found to be performing poorly by at least one of the test statistics we examined. Third, we find that when simple models do not perform well, this appears to be typically a consequence of failing to fully account for heterogeneity in the rate of the evolution. We advocate that assessment of model performance should become a routine component of phylogenetic comparative expression studies; doing so can improve the reliability of inferences and inspire the development of novel models.

Unifying approaches from statistical genetics and phylogenetics for mapping phenotypes in structured populations

“…other studies have shown expression counts for many genes in this dataset (and many others) are not welldescribed by a Brownian process [66,131]. As such, some of their observed correlations could be spurious due to unmodeled phylogenetic structure [35].…”

A tale of too many trees: a conundrum for phylogenetic regression

“…For example, many studies either assume a single species-level phylogeny that has been estimated using coalescent-based (Doña and Johnson, 2023) or traditional concatenation approaches (Hensen et al 2023), or a single gene tree (e.g., Ross et al 2004; Al-Kahtani et al 2004; Kamilar and Cooper 2013; Adams et al 2014; Gu 2016; Dunn et al 2018; Chen et al 2023) to model trait evolution. However, this assumption may (Dimayacyac et al 2023) or may not (Hahn and Nakhleh 2016a) be the best strategy. On the other hand, modeling evolution as a function of a particular gene tree may prove beneficial for traits predicted to exhibit a more direct one-to-one correspondence with a single tree, such as expression of a gene largely regulated by cis elements near its encoded locus (Chen et al 2019; Bertram et al 2022; Bastide et al 2023; Dimayacyac et al 2023).…”

“…However, this assumption may (Dimayacyac et al 2023) or may not (Hahn and Nakhleh 2016a) be the best strategy. On the other hand, modeling evolution as a function of a particular gene tree may prove beneficial for traits predicted to exhibit a more direct one-to-one correspondence with a single tree, such as expression of a gene largely regulated by cis elements near its encoded locus (Chen et al 2019; Bertram et al 2022; Bastide et al 2023; Dimayacyac et al 2023). Perhaps such scenarios arguably represent a best case in which we might at least hope to match tree with trait.…”

“…Perhaps such scenarios arguably represent a best case in which we might at least hope to match tree with trait. Fitting PCMs to gene expression data has also garnered recent considerable interest for modeling functional genomic evolution across cells, tissues, and species (Chen et al 2019; Bertram et al 2022; Adams et al 2023; Bastide et al 2023; Dimayacyac et al 2023). Somewhat surprisingly, a recent study found that modeling gene expression as a function of the overall species tree improved the fit of Ornstein-Uhlenbeck models (Dimayacyac et al 2023), even when compared to local gene trees.…”

“…Fitting PCMs to gene expression data has also garnered recent considerable interest for modeling functional genomic evolution across cells, tissues, and species (Chen et al 2019; Bertram et al 2022; Adams et al 2023; Bastide et al 2023; Dimayacyac et al 2023). Somewhat surprisingly, a recent study found that modeling gene expression as a function of the overall species tree improved the fit of Ornstein-Uhlenbeck models (Dimayacyac et al 2023), even when compared to local gene trees. Other traits, however, may be subject to more complex architectures that incorporate multiple genetic loci, each with their own genealogical history.…”