A basic structural unit in our classification system of local DNA conformations is a dinucleotide. Dinucleotides are classified into conformational classes NtC based on the analysis of their backbone torsions δ, ε, ζ, α1, β1, γ1 and δ1, and glycosidic torsions χ and χ1 (Fig.).

Figure:    Dinucleotide unit of DNA structure with the set of selected torsions.

NtC classes are defined by torsion angle ranges and each class is annotated by a short description describing its main structural features (Tab. 1). PDB files of class representants are available.

While NtC classes are useful for DNA structure fine analysis, a coarser classification is needed to facilitate the human understanding of DNA structure and to better characterize the type of DNA on a local level. Thus, we assembled dinucleotide classes NtC into twelve groups forming a structural alphabet CANA (Tab. 2).

The search field (Fig. A) on the top of the Home page enables a fulltext search in the whole database. Database records can be searched by a PDB ID, author name or keyword. Keyword searches are performed within a structure title, authors and structure publication abstract. The search field is case insensitive. By hitting the Search button, relevant records with a searched term highlighted in red are returned.

Figure:    Fulltext search.

From the Home page, the database can be browsed by its dinucleotide content. Dinucleotide browse can be further restricted by a sequence, NtC or CANA class. Browsing the database by other characteristics (i.e., by structures, chains, residues, base pairs or base steps) is available from a separate Browse section.

DNA structural data are retrieved from the PDB database (http://www.pdb.org) every Wednesday at 10:00 UTC. The date of the last update and a number of structures in the database are given in the bottom of the Home page. The details of the key measures of the database coverage are available from the separate Statistics section.

A Statistics section describes the database coverage in terms of individual structural units: structures, chains, dinucleotides, residues, base pairs and base steps. Data about each structural unit are arranged into a three-boxed layout. A first box contains the depiction of a structural unit (Fig. A), a second box lists the available type of statistics (Fig. B) that is, upon a mouse click, graphed in a third box (Fig. C). The graph of a selected statistic can be printed or saved as an image (Fig. D) in PNG, JPG, PDF or SVG format.

Figure:    Statistics for structures in Dolbico database.

Advanced search enables to build flexible and powerful queries over the Dolbico database. Each query begins by the definition of a searched structural unit (Fig. A). Structural unit search can be made more specific by defining limiting terms (Fig. B). Limiting terms are added using the plus sign (Fig. C) and removed by using the minus sign (Fig. D). A whole search branch can be removed by selecting a "choose ..." option from the first leftmost drop box. If a parent structural unit is defined for a searched structural unit, parent structural unit selector appears (Fig. E) and a search can further be limited by properties of a parent structural unit. This enables, for example, to search all A-form dinucleotides (a searched structural unit) in DNA structures resolved by X-ray with the resolution between 1 Å and 2 Å (Fig.).

Figure:    Advanced search.

The search output is divided into several sections. A header (Fig. A) shows the count of records matching a query followed by three icons enabling the export of results and by an icon turning on and off the display of the statistics of found records. In Dolbico, two following types of results can be exported: (i) details about structures (e.g., PDB ID, structure title, experimental metadata) containing structural units complying with a given query and (ii) details about found structural units. Results are exported as comma-separated values (CSV) files which enables their further analysis in specialized statistical software.

Below a header, a pagination enabling a quick movement between result pages is displayed (Fig. B). Each result page consists of query result panel (Fig. C) and visualization panel (Fig. D). The content of query result panel differs depending on the type of a searched structural unit. For example, if a searched structural unit is a base pair, the following details are shown (Fig. C) in query result panel: structure PDB ID, identification (residue names and IDs) of a base pair complying a query, an indication, if a base pair is classified as Watson-Crick or not and base pair parameters. If a line with one qery result is clicked, a corresponding structural unit is highlighted in red in visualization panel employing the interactive JSmol (http://sourceforge.net/projects/jsmol) applet (Fig. D).

Figure:    Output of search.

The Browse section enables the content of the Dolbico database to be browsed by any type of a structural unit with respect to defined properties (Fig. A). If the "any" value is specified (Fig. B), a given property is ignored.

Figure:    Browse for structures by defined properties.

The Analyse database section enables the quick overview of dinucleotide content in the Dolbico database. The distribution (Fig. E) of individual torsions (Fig. D) can be displayed either for each CANA (Fig. A), or for each NtC (Fig. B) class. If CANA class is selected, its underlying NtC classes are framed in blue (Fig. B). In addition to distribution graph, basic statistics of a selected torsion angle are displayed (Fig. C).

Figure:    Analysis of dinucleotide content of the database.