Using phylogenetically-informed annotation (PIA) to search for light-interacting genes in transcriptomes from non-model organisms

Background: Tools for high throughput sequencing and de novo assembly make the analysis of transcriptomes (i.e. the suite of genes expressed in a tissue) feasible for almost any organism. Yet a challenge for biologists is that it can be difficult to assign identities to gene sequences, especially from non-model organisms. Phylogenetic analyses are one useful method for assigning identities to these sequences, but such methods tend to be time-consuming because of the need to re-calculate trees for every gene of interest and each time a new data set is analyzed. In response, we employed existing tools for phylogenetic analysis to produce a computationally efficient, tree-based approach for annotating transcriptomes or new genomes that we term Phylogenetically-Informed Annotation (PIA), which places uncharacterized genes into pre-calculated phylogenies of gene families. Results: We generated maximum likelihood trees for 109 genes from a Light Interaction Toolkit (LIT), a collection of genes that underlie the function or development of light-interacting structures in metazoans. To do so, we searched protein sequences predicted from 29 fully-sequenced genomes and built trees using tools for phylogenetic analysis in the Osiris package of Galaxy (an open-source workflow management system). Next, to rapidly annotate transcriptomes from organisms that lack sequenced genomes, we repurposed a maximum likelihood-based Evolutionary Placement Algorithm (implemented in RAxML) to place sequences of potential LIT genes on to our pre-calculated gene trees. Finally, we implemented PIA in Galaxy and used it to search for LIT genes in 28 newly-sequenced transcriptomes from the light-interacting tissues of a range of cephalopod mollusks, arthropods, and cubozoan cnidarians. Our new trees for LIT genes are available on the Bitbucket public repository (http://bitbucket.org/osiris_phylogenetics/pia/) and we demonstrate PIA on a publicly-accessible web server (http://galaxy-dev.cnsi.ucsb.edu/pia/). Conclusions: Our new trees for LIT genes will be a valuable resource for researchers studying the evolution of eyes or other light-interacting structures. We also introduce PIA, a high throughput method for using phylogenetic relationships to identify LIT genes in transcriptomes from non-model organisms. With simple modifications, our methods may be used to search for different sets of genes or to annotate data sets from taxa outside of Metazoa.