of a few subfamilies, specifically those bearing leucine-rich repeat domains. Because members of these subfamilies are known to have roles in defence/resistance responses, it has been suggested that the wide expansion was probably a result of adaptation to rapidly evolving pathogens. In contrast, the RLK subfamilies with developmental functions are conserved in size. The expansion of the PK superfamily seems to correlate well with increasing developmental complexity. For example, the green PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19811080 alga Ostreococcus tauri and moss contain 93 and 685 PKs, GFT-505 site respectively, compared with 1008 PKs in Arabidopsis. In addition, there are considerable differences in the sizes of kinomes among plant species by more than 4-fold. However, because the kinome of various plant species constitutes a relatively similar percentage of the whole-genome protein-coding genes, it seems likely that the PK superfamily has undergone similar mechanisms of expansion in flowering plants. Functional characterization studies of PKs in Arabidopsis thaliana during the last two decades have positioned PKs as core components of various signalling pathways controlling multiple biological processes. In soybean, however, only a limited number of PKs have been functionally characterized, despite the recent development of functional genomics tools and a published soybean genome. Soybean PKs have been found to function in the response to biotic stress and abiotic stress, as well as in the equilibrium of gene expression between disease resistance and growth and development. Whole-genome sequencing of several plant species has allowed large-scale identification of plant kinomes with higher resolution than earlier studies. With the availability of the soybean whole-genome sequences along with annotation of the encoded proteins, it is now possible to identify and functionally classify the PK family using large-scale phylogenetic approaches, which was our goal for this 
study. We identified 2166 putative PK genes, which were functionally classified into groups, families, and subfamilies. Our analysis of soybean PKs also included their chromosomal location, gene structure, duplication events, expansion, subcellular localization, expression profiles, and co-expression relationships. Materials and methods Identification and classification of soybean PKs To identify soybean PKs at the genome level, all protein-coding genes were downloaded from the latest version of the soybean genome from Phytozome. Hidden Markov models of the `typical’ Pkinase clan were used to search for putative PKs using HMMER v. 3.0 with an E-value cut-off of <1.0. After this initial screen, 2352 sequences containing the PK domain were identified. Only the longest variant of each gene was retained and all other redundant sequences were deleted. All the remaining sequences were further aligned with the PFAM kinase domain models to confirm the presence of kinase domains and eliminate pseudogenes as described previously by Lehti-Shiu and Shiu. In this analysis, the putative PKs were considered typical PKs only if the domain alignments covered at least 50% of the PFAM domain models. Finally, a total of 2166 proteins containing at least one kinase domain were identified and assigned as soybean `typical' PKs. Classification of identified kinases to groups, families, and subfamilies was defined using HMMs of the different subfamilies developed by Lehti-Shiu and Shiu based on PK sequences obtained from 21 plant species. The classifi