Chapter 3:  Macromolecular Structure Databases

Sources of Primary Data

To build MMDB (1), 3D structure data are retrieved from the PDB database (6) administered by the Research Collaboratory for Structural Bioinformatics (RCSB). In all cases, the structures in MMDB have been determined by experimental methods, primarily X-ray crystallography and Nuclear Magnetic Resonance (NMR) spectroscopy. Theoretical structure models are omitted. The data in each record are then checked for agreement between the atomic coordinates and the primary sequence, and the sequence data are then extracted from the coordinate set. The resulting agreement between sequence and structure allows the record to be linked efficiently into searches and alignment displays involving other NCBI databases.

Annotation of 3D Domains

After initial processing, 3D domains are automatically identified within each MMDB record. 3D domains are annotations on individual MMDB structures that define the boundaries of compact substructures contained within them. In this way, they are similar to secondary structure annotations that define the boundaries of helical or β-strand substructures. Because proteins are often similar at the level of domains, VAST compares each 3D domain to every other one and to complete polypeptide chains. The results are stored in Entrez as a Related 3D Domain link.

To identify 3D domains within a polypeptide chain, MMDB's domain parser searches for one or more breakpoints in the structure. These breakpoints fall between major secondary structure elements such that the ratio of intra- to interdomain contacts remains above a set threshold. The 3D domains identified in this way provide a means to both increase the sensitivity of structure neighbor calculations and also present 3D superpositions based on compact domains as well as on complete polypeptide chains. They are not intended to represent domains identified by comparative sequence and structure analysis, nor do they represent modules that recur in related proteins, although there is often good agreement between domain boundaries identified by these methods.

Links to Other NCBI Resources

After initially processing the PDB record, structure staff add a number of links and other information that further integrate the MMDB record with other NCBI resources. To begin, the sequence information extracted from the PDB record is entered into the Entrez Protein and/or Nucleotide databases as appropriate, providing a means to retrieve the structure information from sequence searches. As with all sequences in Entrez, precomputed BLAST searches are then performed on these sequences, linking them to other molecules of similar sequence. For proteins, these BLAST neighbors may be different than those determined by VAST; whereas VAST uses a conservative significance threshold, the structural similarities it detects often represent remote relationships not detectable by sequence comparison. The literature citations in the PDB record are linked to PubMed so that Entrez searches can allow access to the original descriptions of the structure determinations. Finally, semiautomatic processing of the “source” field of the PDB record provides links to the NCBI Taxonomy database. Although these links normally follow the genus and species information given, in some cases this information is either absent in the PDB record or refers only to how a sample was obtained. In these cases, the staff manually enters the appropriate taxonomy links.

The MMDB Record

Figure 2

The Structure Summary page

Figure 2

.

   The Structure Summary page

The page consists of three parts: the header, the view bar, and the graphic display. The header contains basic identifying information about the record: a description of the protein (Description:), the author list (Deposition:), the species of origin (Taxonomy:), literature references (Reference:), the MMDB-ID (MMDB:), and the PDB code (PDB:). Several of these data serve as links to additional information. For example, the species name links to the Taxonomy browser, the literature references link to PubMed, and the PDB code links to the PDB Web site. The view bar allows the user to view the structure record either as a graphic with Cn3D or as a text record in either ASN.1, PDB (RasMol), or Mage formats. The latter can also be downloaded directly from this page. The graphic display contains a variety of information and links to related databases: (a) The Chain bar. Each chain of the molecule is displayed as a dark bar labeled with residue numbers. To the left of this bar is a Protein hyperlink that takes the user to a view of the protein record in Entrez Protein. The bar itself is also a hyperlink and displays the VAST neighbors of the chain. If a structure contains nucleotide sequences, they are displayed in the order contained in the PDB record. A Nucleotide hyperlink to their left takes the user to the appropriate record in Entrez Nucleotide. (b) The VAST (3D) Domain bar. The colored bars immediately below the Chain bar indicate the locations of structural domains found by the original MMDB processing of the protein. In many cases, such a domain contains unconnected sections of the protein sequence, and in such cases, discontinuous pieces making up the domain will have bars of the same color. To the left of the Domain bar is a 3D Domains hyperlink (3d Domains) that launches the 3D Domains browser in Entrez, where the user can find information about each constituent domain. Selecting a colored segment displays the VAST Structure Neighbors page for that domain. (c) The CD bar. Below the VAST Domain bar are rounded, rectangular bars representing conserved domains found by a CD-Search. The bars identify the best scoring hits; overlapping hits are shown only if the mutual overlap with hits having better scores is less than 50%. The CDs hyperlink to the left of the bar displays the CD records in Entrez Domains. Each of the colored bars is also a hyperlink that displays the corresponding CD Summary page configured to show the multiple alignment of the protein sequence with members of the selected CD.

Figure 2

VAST Structure Neighbors

Figure 3

VAST Structure Neighbors page

Figure 3

.

   VAST Structure Neighbors page

The top portion of the page contains identifying information about the 3D Domain, along with three functional bars. (a) The View bar. This bar allows a user to view a selected alignment either as a graphic using Cn3D or as a sequence alignment in HTML, text, or mFASTA format. The user may select which chains to display in the alignment by checking the boxes that appear to the left of each neighbor in the lower portion of the page. (b) The nr-PDB bar. This bar allows a user to either display all matching records in MMDB or to limit the displayed domains to only representatives of the selected nr-PDB set. The user may also select how the matching domains are sorted in the display and whether the results are shown as graphics or as tabulated data. (c) The Find bar. This bar allows the user to find specific structural neighbors by entering their PDB or MMDB identifiers. (d) The lower portion of the page displays a graphical alignment of the various matching domains. The upper three bars show summary information about the query sequence: the top bar shows the maximum extent of alignment found on all the sequences displayed on the current page (users should note that the appearance of this bar, therefore, depends on which hits are displayed); the middle bar represents the query sequence itself that served as input for the VAST search; and the lower bar shows any matching CDs and is identical to the CD bar on the Structure Summary page. Listed below these three summary bars are the hits from the VAST search, sorted according to the selection in the nr-PDB bar. Aligned regions are shown in red, with gaps indicating unaligned regions. To the left of each domain accession is a check box that can be used to select any combination of domains to be displayed either on this page or using Cn3D. Moreover, each of the bars in the display is itself a link, and placing the mouse pointer over any bar reveals both the extent of the alignment by residue number and the data linked to the bar.

Figure 3

Figure 2

nr-PDB

The non-redundant PDB database (nr-PDB) is a collection of four sets of sequence-dissimilar cluster PDB polypeptide chains assembled by NCBI Structure staff. The four sets differ only in their respective levels of non-redundancy. The staff assembles each set by comparing all the chains available from PDB with each other using the BLAST algorithm. The chains are then clustered into groups of similar sequence using a single-linkage clustering procedure. Chains within a sequence-similar group are automatically ranked according to the quality of their structural data. Details of the measures used to determine structure precision and completeness and the methodology of assembling the nr-PDB clusters can be found on the nr-PDB Web page.

What Is a Conserved Domain (CD)?

CDs are recurring units in polypeptide chains (sequence and structure motifs), the extents of which can be determined by comparative analysis. Molecular evolution uses such domains as building blocks and these may be recombined in different arrangements to make different proteins with different functions. The CDD contains sequence alignments that define the features that are conserved within each domain family. Therefore, the CDD serves as a classification resource that groups proteins based on the presence of these predefined domains. CDD entries often name the domain family and describe the role of conserved residues in binding or catalysis. Conserved domains are displayed in MMDB Structure summaries and link to a sequence alignment showing other proteins in which the domain is conserved, which may provide clues on protein function.

Sources of Primary Data

The collections of domain alignments in the CDD are imported either from two databases outside of the NCBI, named Pfam (8) and SMART (9); from the NCBI COB database; from another NCBI collection named LOAD; and from a database curated by the CDD staff. The first task is to identify the underlying sequences in each collection and then link these sequences to the corresponding ones in Entrez. If the CDD staff cannot find the Accession numbers for the sequences in the records from the source databases, they locate appropriate sequences using BLAST. Particular attention is paid to any resulting match that is linked to a structure record in MMDB, and the staff substitute alignment rows with such sequences whenever possible. After the staff imports a collection, they then choose a sequence that best represents the family. Whenever possible, the staff chooses a representative that has a structure record in MMDB.

The Position-specific Score Matrix (PSSM)

Once imported and constructed, each domain alignment in CDD is used to calculate a model sequence, called a consensus sequence, for each CD. The consensus sequence lists the most frequently found residue in each position in the alignment; however, for a sequence position to be included in the consensus sequence, it must be present in at least 50% of the aligned sequences. Aligned columns covered by the consensus sequence are then used to calculate a PSSM, which memorizes the degree to which particular residues are conserved at each position in the sequence. Once calculated, the PSSM is stored with the alignment and becomes part of the CDD. The RPS-BLAST tool locates CDs within a query sequence by searching against this database of PSSMs.

Reverse Position-specific BLAST (RPS-BLAST)

RPS-BLAST (Chapter 16) is a variant of the popular Position-specific Iterated BLAST (PSI-BLAST) program. PSI-BLAST finds sequences similar to the query and uses the resulting alignments to build a PSSM for the query. With this PSSM the database is scanned again to draw in more hits and further refine the scoring model. RPS-BLAST uses a query sequence to search a database of precalculated PSSMs and report significant hits in a single pass. The role of the PSSM has changed from “query” to “subject”; hence, the term “reverse” in RPS-BLAST. RPS-BLAST is the search tool used in the CD-Search service.

The CD Summary

Figure 4

CD summary page

Figure 4

.

   CD summary page

The top of the page serves as a header and reports a variety of identifying information, including the name and description of the CD, other related CDs with links to their summary pages, as well as the source database, status, and creation date of the CD. A taxonomic node link (Taxa:) launches the Taxonomy Browser, whereas a Proteins link (Proteins:) uses CDART to show other proteins that contain the CD. Below the header is the interface for viewing the CD alignment, which can be done either graphically with Cn3D (if the CD contains a sequence with structural data) or in HTML, text, or mFASTA format. It is also possible to view a selected number of the top-listed sequences, sequences from the most diverse members, or sequences most similar to the query. In addition, users may now select sequences with the NCBI Taxonomy Common Tree tool. The lower portion of the page contains the alignment itself. Members with a structural record in MMDB are listed first, and the identifier of each sequence links to the corresponding record.

Figure 4

CD Records Curated at NCBI

Box 3

Example query: finding and viewing CDs in a protein

(more...)

Figure 7

Sequence and structure views of the TOP1Ac conserved (more...)

Figure 7

.

   Sequence and structure views of the TOP1Ac conserved domain common to type III bacterial and eukaryotic DNA topoisomerases

The upper window displays the structure of the domain with the residues colored according to their sequence conservation, with red indicating high conservation and blue indicating low conservation. The nucleotide bound at site II is shown as an orange space-filling model, and the residues involved in this binding site are yellow. The lower window displays the sequence alignment for the domain with aligned residues shown as colored capital letters. Residues aligned to three of the binding site residues are highlighted in yellow. The sequence for NP_004609 (gi 10835218) occupies the bottom row.

Figure 7

The Distinction between 3D Domains and CDs

The term “domain” refers in general to a distinct functional and/or structural unit of a protein. Each polypeptide chain in MMDB is analyzed for the presence of two classes of domains, and it is important for users to understand the difference between them. One class, called 3D Domains, is based solely on similar, compact substructures, whereas the second class, called Conserved Domains (CDs), is based solely on conserved sequence motifs. These two classifications often agree, because the compact substructures within a protein often correspond to domains joined by recombination in the evolutionary history of a protein. Note that CD links can be identified even when no 3D structures within a family are known. Moreover, 3D Domain links may also indicate relationships either to structures not included in CDD entries or to structures so distantly related that no significant similarity can be found by sequence comparisons.

Why Would I Want to Do This?

• To determine the overall shape and size of a protein

• To locate a residue of interest in the overall structure

• To locate residues in close proximity to a residue of interest

• To develop or test chemical hypotheses regarding an enzyme mechanism

• To locate or predict possible binding sites of a ligand

• To interpret mutation studies

• To find areas of positive or negative charge on the protein surface

• To locate particularly hydrophobic or hydrophilic regions of a protein

• To infer the 3D structure and related properties of a protein with unknown structure from the structure of a homologous protein

• To study evolutionary processes at the level of molecular structure

• To study the function of a protein

• To study the molecular basis of disease and design novel treatments

Alternative Points of Entry

Structure Summary pages can also be found from the following NCBI databases and tools:

• Select the Structure links to the right of any Entrez record found; records with Structure links can also be located by choosing Structure links from the Display pull-down menu.

• Select the Related Sequences link to the right of an Entrez record to find proteins related by sequence similarity and then select Structure links in the Display pull-down menu.

• Choose the PDB database from a blastp (protein-protein BLAST) search; only sequences with structure records will be retrieved by BLAST. The Related Structures link provides 3D views in Cn3D.

• Select the 3D Structures button on any BLink report to show those BLAST hits for which structural data are available.

• From the results of any protein BLAST search, click on a red 'S' linkout to view the sequence alignment with a structure record.

Viewing 3D Structures

Why Would I Want to Do This?

• To determine structurally conserved regions in a protein family

• To locate the structural equivalent of a residue of interest in another related protein

• To gain insights into the allowable structural variability in a particular protein family

• To develop or test chemical hypotheses regarding an enzyme mechanism

• To predict possible binding sites of a ligand from the location of a binding site in a related protein

• To identify sites where conformational changes are concentrated

• To interpret mutation studies

• To find areas of conserved positive or negative charge on the protein surface

• To locate conserved hydrophobic or hydrophilic regions of a protein

• To identify evolutionary relationships across protein families

• To identify functionally equivalent proteins with little or no sequence conservation

How to Begin

Figure 2

• The 3D Domains link transfers the user to Entrez 3D Domains, showing a list of the VAST neighbors.

• Selecting the chain bar displays the VAST Structure Neighbors page for the entire chain.

• Selecting a 3D Domain bar displays the VAST Structure Neighbors page for the selected domain.

Alternative Points of Entry

• From any Entrez search, select Related 3D Domains to the right of any record found to view the Vast Structure Neighbors page.

Why Would I Want to Do This?

• To locate functional domains within a protein

• To predict the function of a protein whose function is unknown

• To establish evolutionary relationships across protein families

• To interpret mutation studies

• To predict the structure of a protein of unknown structure

How to Begin

Following the Domains link for any protein in Entrez, one can find the conserved domains within that protein. The CD-Search (or Protein BLAST, with CD-Search option selected) can be used to find conserved domains (CDs) within a protein. Either the Accession number, gi number, or the FASTA sequence can be used as a query.

Alternative Points of Entry

Information on the CDs contained within a protein can also be found from these databases and tools:

• From any Entrez search: select the Domains link to the right of a displayed record.

• From the Structure Summary page of a MMDB record: this page displays the CDs within each protein chain immediately below the 3D Domain bar in the graphic display. Selecting the CDs link shows the CD-Search results page.

• From an Entrez Domains search: choose Domains from the Entrez Search pull-down menu and enter a search term to retrieve a list of CDs. Clicking on any resulting CD displays the CD Summary page. To find the location of this CD in an aligned protein, select the CD link following a protein name in the bottom portion of this page.

• From the CDD page: locate CDs by entering text terms into the search box and proceed as for an Entrez CD search.

• From a BLink report: select the CDD-Search button to display the CD-Search results page.

• From the BLAST main page: follow the RPS-BLAST link to load the CD-Search page.

Viewing Conserved Domains

Results from a CD search are displayed as colored bars underneath a sequence ruler. Moving the mouse over these bars reveals the identity of each domain; domains are also listed in a format similar to BLAST summary output (Chapter 16). Pairwise alignments between the matched region of the target protein and the representative sequence of each domain are shown below the bar. Red letters indicate residues identical to those in the representative sequence, whereas blue letters indicate residues with a positive BLOSUM62 score in the BLAST alignment.

Viewing Multiple Alignments of a Query Protein with Members of a Conserved Domain

These can be displayed by clicking a CD bar within a MMDB Structure Summary page or from a hyperlinked CD name on a CD-Search results page.

Viewing CD Alignments in the Context of 3D Structure

If members of a CD have MMDB records, one of these records can be viewed as a 3D image along with the sequence alignment using Cn3D (launched by selecting the pink dot on a CD-Search results page). As in other alignment views, colored capital letters indicate aligned residues, allowing the sequence of the protein sequence of interest to be mapped onto the available 3D structure.

Why Would I Want to Do This?

• To locate related functional domains in other protein families

• To gain insights into how a given CD is situated within a protein relative to other CDs

• To explore functional links between different CDs

• To predict the function of a protein whose function is unknown

• To establish evolutionary relationships across protein families

How to Begin

Following the Domain Relatives link for any protein in Entrez, one can find other proteins with similar domain architecture. The Conserved Domain Architecture Retrieval Tool (CDART) can take an Accession number or the FASTA sequence as a query to find out the domain architecture of a protein sequence and list other proteins with related domain architectures.

Alternative Points of Entry

• From a CD-Search results page, click Show Domain Relatives

• From a CD-Summary page, click the Proteins link

• From an Entrez Domains searc, click the Proteins link in the Links menu

Results of a CDART Search

Figure 5

A CDART results page

Figure 5

.

   A CDART results page

At the top of the CDART results page in a yellow box, the query sequence CDs are represented as “beads on a string”. Each CD had a unique color and shape and is labeled both in the display itself and in a legend located at the bottom of the page. The shapes representing CDs are hyperlinked to the corresponding CD summary page. The matching proteins to the query are listed below the yellow box, ranked according to the number of non-redundant hits to the domains in the query sequence. Each match is either a single protein, in which case its Accession number is shown, or is a cluster of very similar proteins, in which case the number of members in the cluster is shown. Cluster members can be displayed by selecting the logo to the left of its diagram. Selecting any protein Accession number displays the flatfile for that protein. To the right of any drawing for a single protein (either on the main results page or after expanding a protein cluster) is a more> link, which displays the CD-Search results page for the selected protein so that the sequence alignment, e.g., of a CDART hit with a CD contained in the original protein of interest, can be examined.

Figure 5

Integration with Other NCBI Resources

As illustrated in the sections above, there are numerous connections between the Structure resources and other databases and tools available at the NCBI. What follows is a listing of major tools that support connections.

Links to Non-NCBI Resources

Exporting Graphics Files from Cn3D

Structures displayed in Cn3D can be exported as a Portable Network Graphics (PNG) file from within Cn3D (the Export PNG command in the File menu). The structure file itself, in the orientation currently being viewed, can also be saved for later launching in Cn3D.

Saving Individual MMDB Records

Individual MMDB records can be saved/downloaded to a local computer directly from the Structure Summary page for that record. Save File in the View bar downloads the file in a choice of three formats: ASN.1 (select Cn3D); PDB (select RasMol); or Mage (select Mage).

Saving VAST Alignments

Alignments of VAST neighbors can be saved/downloaded from the VAST Structure Neighbors page of any MMDB record. By selecting options in the View Alignment pull-down menu, the alignment data can be saved, formatted as HTML, text, or mFASTA, and then saved.

MMDB

Users can download the NCBI Structure databases from the NCBI FTP site: ftp://ftp.ncbi.nih.gov/mmdb. A Readme file contains descriptions of the contents and information about recent updates. Within the mmdb directory are four subdirectories that contain the following data:

• mmdbdata: the current MMDB database (NOTE: these files can not be read directly by Cn3D).

• vastdata: the current set of VAST neighbor annotations to MMDB records

• nrtable: the current non-redundant PDB database

• pdbeast: table listing the taxonomic classification of MMDB records

CDD

CDD data can be downloaded from ftp://ftp.ncbi.nih.gov/pub/mmdb/cdd. A Readme file contains descriptions of the data archives. Users can download the PSSMs for each CD record, the sequence alignments in mFASTA format, or a text file containing the accessions and descriptions of all CD records.