Beyond gene ontology (GO): using biocuration approach to improve the gene nomenclature and functional annotation of rice S-domain kinase subfamily

The S-domain subfamily of receptor-like kinases (SDRLKs) in plants is poorly characterized. Most members of this subfamily are currently assigned gene function based on the <i>S</i>-locus Receptor Kinase from Brassica that acts as the female determinant of self-incompatibility (SI). However, Brassica like SI mechanisms does not exist in most plants. Thus, automated Gene Ontology (GO) pipelines are not sufficient for functional annotation of SDRLK subfamily members and lead to erroneous association with the GO biological process of SI. Here, we show that manual bio-curation can help to correct and improve the gene annotations and association with relevant biological processes. Using publicly available genomic and transcriptome datasets, we conducted a detailed analysis of the expansion of the rice (<i>Oryza sativa</i>) SDRLK subfamily, the structure of individual genes and proteins, and their expression.The 144-member SDRLK family in rice consists of 82 receptor-like kinases (RLKs) (67 full-length, 15 truncated),12 receptor-like proteins, 14 SD kinases, 26 kinase-like and 10 GnK2 domain-containing kinases and RLKs. Except for nine genes, all other SDRLK family members are transcribed in rice, but they vary in their tissue-specific and stress-response expression profiles. Furthermore, 98 genes show differential expression under biotic stress and 98 genes show differential expression under abiotic stress conditions, but share 81 genes in common.Our analysis led to the identification of candidate genes likely to play important roles in plant development, pathogen resistance, and abiotic stress tolerance. We propose a nomenclature for 144 SDRLK gene family members based on gene/protein conserved structural features, gene expression profiles, and literature review. Our biocuration approach, rooted in the principles of findability, accessibility, interoperability and reusability, sets forth an example of how manual annotation of large-gene families can fill in the knowledge gap that exists due to the implementation of automated GO projections, thereby helping to improve the quality and contents of public databases.