GRASSIUS: a platform for comparative regulatory genomics across the grasses

Abstract

Transcription factors (TFs) are major players in gene regulatory networks and interactions between TFs and their target genes furnish spatiotemporal patterns of gene expression. Establishing the architecture of regulatory networks requires gathering information on TFs, their targets in the genome, and the corresponding binding sites. We have developed GRASSIUS (Grass Regulatory Information Services) as a knowledge-based Web resource that integrates information on TFs and gene promoters across the grasses. In its initial implementation, GRASSIUS consists of two separate, yet linked, databases. GrassTFDB holds information on TFs from maize (Zea mays), sorghum (Sorghum bicolor), sugarcane (Saccharum spp.), and rice (Oryza sativa). TFs are classified into families and phylogenetic relationships begin to uncover orthologous relationships among the participating species. This database also provides a centralized clearinghouse for TF synonyms in the grasses. GrassTFDB is linked to the grass TFome collection, which provides clones in recombination-based vectors corresponding to full-length open reading frames for a growing number of grass TFs. GrassPROMDB contains promoter and cis-regulatory element information for those grass species and genes for which enough data are available. The integration of GrassTFDB and GrassPROMDB will be accomplished through GrassRegNet as a first step in representing the architecture of grass regulatory networks. GRASSIUS can be accessed from www.grassius.org.

Figure 1.

Estimation of TF numbers in grass genomes. Correlation between the number of TFs and the total number of genes in genomes was based on completely annotated plant genomes of Arabidopsis, rice, poplar, and Chlamy (black circles). A best-fit linear regression (r² = 0.87) was used to estimate the predicted number of TFs in maize, sorghum, and sugarcane (Table I). A similar analysis was conducted for nonplant organisms (blue triangles), including yeast (Saccharomyces cerevisiae), Caenorhabditis elegans, fruitfly (Drosophila melanogaster), mouse (Mus musculus), and human (Homo sapiens), and a best-fit linear regression line was drawn (r² = 0.74; blue dashed line).

Figure 2.

Flow diagram describing the steps involved in the generation of GrassTFDB. Details available in “Materials and Methods.”

Figure 3.

Screen shot showing query possibilities for the GrassTFDB database of GRASSIUS. A, All families in a species or members of a single family can be retrieved by clicking the species name or selecting the family name from the pull-down menu, respectively. B, Specific TFs or families can be retrieved by performing searches by selecting a particular family, or by keywords. Multiple TFs can be searched simultaneously by using the batch search option. C, BLAST application allows TFs to be searched in GrassTFDB by protein (blastp from the pull-down menu) or DNA (blastn from the pull-down menu) sequence. D, Phylogenetic trees of TF families can be retrieved by selecting the families.

Figure 4.

Screen shot of a part of the maize MYB family query result. A short description of the family is provided, including one or more key references. The first column indicates the name of the TF, following the guidelines provided (see Letter to the Editor, this issue [Gray et al., 2009]). Names in blue correspond to those accepted, those in gray correspond to those suggested, waiting comments by the community. The protein names provide clickable links to the general information page for each TF (Fig. 6). The Synonym/Gene Name column provides alternate names by which the TF (or the gene encoding it) is known. Fields in blue indicate hyperlinks to other databases (such as, for example, MaizeGDB for maize). The Gene Id column provides links to species-specific external databases. If a clone is available for a particular TF in the TFome collection, or if direct targets for a TF are known, the corresponding columns provide links to the corresponding pages.

Figure 5.

Structure of GrassTFDB. Interconnected MySQL tables contain data for each TF and related TFome clones and binding sites. TF information submitted by the community is stored in the uploads table and integrated into GrassTFDB after review.

Figure 6.

TF general information page. Screen shot of a sample general information page for OsMYB1. The domain structure view is generated using the Bio∷Graphics module, which parses InterProScan results. Each box corresponds to an InterProScan hit and provides a database-specific identification number above and a description below. The name of the database from which the information is retrieved is provided on the left side of the image. Both nucleotide and peptide sequences of the TF are provided in scrollable windows at the bottom of the page. This information page will be expanded as more information on TFs becomes available.