Integrating gene annotation with orthology inference at scale

Abstract

Annotating coding genes and inferring orthologs are two classical challenges in genomics and evolutionary biology that have traditionally been approached separately, which limits scalability. We present TOGA, the first method that integrates gene annotation and orthology inference. TOGA implements a novel paradigm to infer orthologous genes, improves ortholog detection and annotation completeness compared to state-of-the-art methods, and handles even highly-fragmented assemblies. TOGA scales to hundreds of genomes, which we demonstrate by applying it to 488 placental mammal and 308 bird assemblies, creating the largest comparative gene resources so far. Additionally, TOGA detects gene losses, enables selection screens, and automatically provides a superior measure of mammalian genome quality. Together, TOGA is a powerful and scalable method to annotate and compare genes in the genomic era.

Vocal learning-associated convergent evolution in mammalian proteins and regulatory elements

“…We sought to evaluate the presence of convergent genomic specializations between vocal learning mammals using new datasets and computational approaches, focusing on bats as an attractive mammalian model of vocal complexity (6)(7)(8)(9)(10)(11)(12)(13). Specifically, we used protein-coding sequences from genomes generated by the Zoonomia Consortium (14,15) and models of evolutionary rate convergence (16)to identify genes repeatedly associated with the evolution of vocal learning. Motivated by the finding of protein-level convergence, we next profiled open chromatin specializations of multiple brain regions and somatic tissues in the Egyptian fruit bat, a bat with robust vocal plasticity (10)(11)(12), to identify vocalization-associated epigenomic specializations.…”

“…To explore the possibility of shared genomic specializations associated with vocal learning, we first used new protein-coding alignments for hundreds of mammals ( 14 ) to identify genes whose rates of evolution differs between vocal learners and other mammals, and which may thus be under evolutionary selection related to vocal learning ( 16 ). We analyzed 16,209 high-quality gene alignments across 175 boreoeutherian mammals, including 25 vocal learning species (Materials and Methods).…”

A haplotype-resolved genome assembly of the Nile rat facilitates exploration of the genetic basis of diabetes

“…We used TOGA (Tool to infer Orthologs from Genome Alignments) [34] with human and mouse as references to annotate genes in the Nile rat genome. In addition to providing gene annotations, TOGA distinguishes between intact genes and genes with missing sequences or inactivating mutations, which can be used to evaluate the quality of genome assemblies.…”

“…We used TOGA [34] to project protein coding genes from human and house mouse to the Nile rat genome. Overall, 99.7% of TOGA annotated genes in the paternal assembly are also annotated in the maternal assembly using mouse gene models; when using human gene models the number is 96.6%.…”

A high-quality blue whale genome, segmental duplications, and historical demography

“…We applied TOGA using the human GRChg38 assembly and the human GENCODE V38 gene annotation as reference to genomes of cetaceans (query species) (91). Briefly, TOGA uses pairwise genome alignment chains and machine learning to infer orthologous loci for each transcript in the reference annotation, utilizing that orthologous genes exhibit alignments in intronic and flanking intergenic regions.…”