Stage-specific transcriptional signal values were imported from genome-wide microarray analyses of An. gambiae larvae, male sugar-fed adults, female sugar-fed adults, and female blood-fed adults 3, 24, 48, 72, 96 hours and 15 days after a bloodmeal using Affymetrix GCOS software. Values from tissue-specific microarray analyses also were imported using GCOS to quantify genome-wide transcription in fat bodies, midgut, and ovaries at 24 hours after bloodfeeding 34. Functional gene annotation was imported from the Ano-Xcel database 5 to populate angaGEDUCI with keywords and annotation from the ENSEMBL, NCBI non-redundant, GO, PFAM, and SMART databases. Promoter sequences were selected as regions 1.5 kilobases (kb) in length adjacent to the 5'-ends of transcription start sites of genes using genomic data from ENSEMBL (Assembly: AgamP3, Feb 2006; Genebuild: VectorBase, Feb 2006; Database version: 37.3). Transcription factor binding sites from several classes of organisms were imported from the Transcription Factors Database (TFD) available publicly at ftp://ftp.ncbi.nih.gov/repository/TFD/datasets/. Of the 7,066 sites listed in TFD, 6639 (94.0%) are eight nucleotides or longer and 623 (8.82%) contain degenerate notation. Five-hundred and eleven sites in the database were identified in insects (7.23%), of which 499 (97.7%) are eight nucleotides or longer, and 34 (6.65%) contain degeneracy.

The data have been stored as a MySQL relational database that is accessible directly through an Apache web server. A web-based data mining interface is used to manage queries to identify genes that meet specific expression, keyword, and sequence criteria (Figure 1). A sequence comparison program based on the Boyer-Moore algorithm 6 is built into the data-mining interface for comparison of promoter regions of genes within a selected gene set.

The main page of the database provides hyperlinks to: Filter Database, Import Gene Set, Download Data, View Database, Submit Study, Documentation, and Contact. Selection of the Filter Database link opens the data-mining interface and allows users to focus on specific genes that satisfy input criteria based on: 1) stage- and tissue-specific expression, 2) annotated keywords, 3) DNA sequences present in promoter, 3' untranslated regions (UTR), or coding regions, or 4) presence of specific transcription factor binding sites (Figure 1). Queries are conducted by stepwise entry of input criteria with each query imposed on the previous so that all genes currently displayed meet all preceding query criteria as well as the criterion that was last entered. Once a gene set of interest has been selected, users then can use the analysis menu in the interface to search for conserved DNA motifs within the promoters of the gene set, view expression profiles, build a distribution of annotated keywords, or export the set for future retrieval (Figure 2). Detailed annotation and expression data for each gene also can be viewed at any time by selecting the gene identifier link to invoke the description of a gene entry.

Each gene has a corresponding data page that can be accessed by selecting the gene identifier link during data retrieval. Gene entry pages display data from microarray expression analyses for stage- and tissue-specific expression and functional annotation as gathered by Ano-Xcel from ENSEMBL, NCBI non-redundant, GO, PFAM, and SMART databases (Figure 3). A link to the Vectorbase database that contains additional, centralized gene data also is provided on each entry page. User-contributed notes and a form for sharing notes for a gene entry are found below the annotation of each gene. To encourage data sharing, note submission does not require user pre-registration.

After clustering genes into gene sets that show similar patterns of expression, the data-mining interface analysis menu can be used to search for common DNA motifs that may act as regulatory sequences in coordinating these expression patterns. Two parameters must be selected to begin the analysis: 1) motif match length: the desired conserved sequence motif length to search for in the analysis, 2) mismatches: the number of base mismatches allowed between two nearly-conserved sequence motifs without disqualification.

The resulting output from the analysis contains three parts. First, a comparison matrix is displayed indicating the number of conserved motifs found in each pair-wise comparison among every gene in the gene set (Figure 4). Each link in the matrix invokes a new page that prints the promoter sequences of the two genes being compared with areas of sequence conservation and transcription factor binding sites highlighted (Figure 5). Second, a table of the conserved motifs is displayed that compares the frequency of occurrence of each conserved motif within the gene set against the frequency of each motif in all 1) exons, 2) exons and introns, and 3) promoters within the An. gambiae genome (Figure 6). Each motif that matches or contains a transcription factor binding site is indicated in the same output. The third item displayed is a table indicating the frequency of occurrence of each transcription factor binding site of any size found within the gene set (Figure 7). Due to the degeneracy and varied size of transcription factor binding sites in the TFD database, the frequencies reported here are noticeably higher in this item compared to the frequencies in the conserved motif table that precedes it.

The transcription profiles for a gene set can be viewed in batch by using the analysis menu from the data-mining interface after a gene set has been selected. The resulting graphs print transcriptional expression according to developmental stage: larvae, male sugar-fed adults, female sugar-fed adults, and female blood-fed adults 3, 24, 48, 72, 96 hours and 15 days after a bloodmeal (Figure 8).

A keyword distribution listing all keywords found in a gene set, as gathered by Ano-Xcel 5, and their respective frequency of occurrence, can be constructed by using the analysis menu from the data-mining interface (Figure 9).

A gene set can be imported by entering a list of gene identifiers in ENSANGG, ENSANGP, ENSANGT, Probeset ID, or Celera form, or by choosing from a list of pre-defined gene sets. Pre-defined gene sets consist of groups of genes that have been linked to similar function or regulation in existing literature (Figure 10). Users can submit gene sets for automatic and immediate listing as a pre-defined gene set from the same page. Gene sets can be exported from the data-mining interface by using the analysis menu.

The angaGEDUCI database has the capacity to store and integrate additional Affymetrix microarray studies that examine gene expression in An. gambiae. The Submit Study link provides a short form for uploading microarray data and specifications.