Detecting gene innovations for phenotypic diversity across multiple genomes

Gene innovation is a key mechanism on the evolution and phenotypic diversity of life forms. There is a need for tools able to study gene innovation across an increasingly large number of genomic sequences to maximally capitalise our understanding of biological systems. Here we present Comparative-Phylostratigraphy, an open-source software suite that enables to time the emergence of new genes across evolutionary time and to correlate patterns of gene emergence with species traits simultaneously across whole genomes from multiple species. Such a comparative strategy is a new powerful tool for starting to dissect the relationship between gene innovation and phenotypic diversity. We describe and showcase our method by analysing recently published ant genomes. This new methodology identified significant bouts of new gene evolution in ant clades, that are associated with shifts in life-history traits. Our method allows easy integration of new genomic data as it becomes available, and thus will be a valuable analytical tool for evolutionary biologists interested in explaining the evolution of diversity of life at the level of the genes.

💡 Research Summary

The paper introduces Comparative‑Phylostratigraphy, an open‑source software suite designed to map the temporal emergence of new genes across multiple species and to statistically associate these gene‑birth events with phenotypic traits. Traditional phylostratigraphy estimates the age of genes in a single genome by aligning protein sequences against a broad database and assigning each gene to a “stratum” that corresponds to the most ancient common ancestor detectable for that sequence. However, this approach does not readily allow direct comparison among many species, limiting its utility in the era of rapidly expanding genomic resources.

Comparative‑Phylostratigraphy extends the concept by processing whole‑genome protein sets from dozens or hundreds of species in a single, unified pipeline. The workflow consists of four main stages: (1) data preparation, where protein FASTA files and a trait matrix (e.g., life‑history, ecological, or behavioral variables) are supplied; (2) similarity search, performed with DIAMOND (or optionally BLASTp) against a comprehensive reference database such as NCBI nr, followed by filtering based on e‑value and alignment coverage; (3) age assignment, where each protein is linked to the deepest node in a user‑provided, time‑calibrated taxonomy (derived from NCBI Taxonomy) that still yields a significant hit, thereby placing the gene into a specific phylostratum; and (4) statistical integration, which uses generalized linear models, Bayesian hierarchical models, and Poisson‑based change‑point detection to test for correlations between bursts of gene birth and the supplied phenotypic variables. Multiple‑testing correction is applied via the Benjamini‑Hochberg false discovery rate procedure.

The software is written in Python 3, packaged with Conda, and distributed as a Docker image to ensure reproducibility. Output includes a gene‑to‑stratum mapping table, per‑stratum counts of newly originated genes, visualizations such as time‑series plots of gene emergence, heat‑maps linking strata to traits, and functional enrichment reports (GO, KEGG) for genes that appear in each burst.

To demonstrate the method, the authors re‑analyzed 19 publicly available ant genomes spanning a range of social structures, diets, and ecological niches. Their comparative analysis uncovered a pronounced peak of gene emergence around 30–40 million years ago, coinciding with the inferred transition from solitary to complex eusocial colonies in ants. Genes involved in chemical communication, sensory perception, and nutrient metabolism were over‑represented in this burst, suggesting that novel molecular functions facilitated the evolution of sophisticated colony organization. Moreover, a secondary, more recent wave of gene birth (10–20 million years ago) correlated with larger colony sizes and specialized caste systems, indicating ongoing genetic innovation linked to social complexity. Statistical tests confirmed that the number of newly originated genes per stratum was significantly associated with specific life‑history traits after correcting for phylogenetic relatedness.

The authors acknowledge several limitations. The accuracy of stratum assignment depends on the quality and completeness of the reference taxonomy and divergence time estimates; mis‑placed nodes can shift age estimates. Highly conserved genes may be mis‑classified as “old” even when they have undergone functional innovation, while rapidly evolving genes may be under‑detected due to alignment thresholds. The current implementation focuses on protein‑coding sequences, leaving non‑coding RNAs, regulatory elements, and epigenetic modifications outside its scope.

Future directions include integrating transcriptomic, methylation, and chromatin‑accessibility data to capture regulatory innovation, employing machine‑learning models to predict stratum assignments when phylogenetic information is sparse, and scaling the pipeline to handle thousands of genomes as more data become available.

In summary, Comparative‑Phylostratigraphy provides a robust, scalable, and reproducible framework for linking gene‑level evolutionary events to organismal phenotypes across many species. By enabling simultaneous temporal mapping of gene birth and statistical testing against trait data, it opens new avenues for dissecting the genetic underpinnings of phenotypic diversity and for generating testable hypotheses about how bursts of gene innovation drive major evolutionary transitions.