Supplementary Info

Figure 1

A typical Pfam family page. At the top of every Pfam page are six links (Home, Search, Browse, FTP, Help and About), each of which is described in the main text. The "Jump to" box and the "Keyword" search are available on most Pfam pages. The context-specific icons at the top of the page indicate the number of different architectures, the number of sequence segments in the full alignment, the number of documented interactions, the number of different species represented in the full alignment and the number of PDB chain IDs that can be displayed for the family concerned. Each tab gives access to different data, details of which can be found in the main text. As illustrated here for the Piwi family (Pfam accession PF02171), this "Summary" tab displays the Wikipedia article. The "Clans" tab is not grayed out, indicating that Piwi is part of a clan.

• Full figure and legend (104K)
• Figures & Tables index

Family pages

The majority of the information in the website is organized around the family pages where all data for a Pfam family are displayed (Figure 1). Entry via the Pfam identifier or accession number, or clicking on a domain name or graphic anywhere on the website, takes the user to the "Summary" tab for that family. The user is able to navigate through several tabs on the left-hand side of the page, each providing data about some particular aspect of the family: domain organization or architectures, clan, alignments, HMM logo, trees, curation and model, species distribution, interactions and structures. The information in each of these tabs is described below.

Summary: The "Summary" tab is the front-page view, or tab, for the family. Where a Wikipedia article has been written for this family then we display that Wikipedia entry by default. Three tabs are provided at the top of the section (Figure 1), to call up any one of the three possible sets of annotations: Wikipedia, Pfam, or InterPro.

In the absence of a Wikipedia article, the "Pfam" tab is displayed and will start with a brief summary of the function, if known. Sectional headings underneath may include an abstract, "Literature references," clan details and "External database" links. Where a three-dimensional structure for a member of the family exists, a static image of the structure is shown on this page; when multiple structures exist, the user can choose which structure image to show via a drop-down menu (Figure 1).

Users can click on the "InterPro annotation" tab to find the "InterPro entry," and "Gene Ontology" data if available.

Domain organization: This tab shows all the other domains that co-occur on a protein with the domain of interest. In Figure 10 the domain organization of the PepSY family (Pfam accession PF03413) is shown. The PepSY domain is rendered in green and other associated domains in other colors. Hovering the mouse over (mousing over) the domains pops up a tool tip with summary data. Beneath each graphic is a "Show" button that will toggle between the single example and the full set of all instances of that particular organization. This can be useful in determining the relative distances between the domains on the different sequences, how the different sequences might vary in overall length and whether or not any Pfam-Bs co-occur.

Clan: The "Clan" tab gives the name and accession number of the clan to which the family belongs, along with a list of all families in the clan. Clicking on the name or accession number of the clan brings up the full Clan page; see below.

Alignments: The Alignments" tab offers different options for viewing the seed: full (UniProtKB), NCBI and metagenomic alignments. The alignments can be viewed in the following formats: Jalview (available for all alignments), HTML (available for all alignments), Pfam viewer (available for seed and full alignments) or heat map view (available for Pfam-A full alignment only).

Where structures have been solved for members of a family, the HTML view of both the seed and the full alignments will show the consensus secondary structure (SS) (this is the consensus of all the structures in the family) marked below each residue in the relevant sequences. Figure 11 shows the SS lines in the seed HTML alignment for the Pfam domain 3-HAO (Pfam accession PF06052). Keys are also given in the figure for interpreting both the secondary-structure symbols and the ClustalX colorings.⁶

In the heat map view (Figure 12), the color of each residue reflects the alignment uncertainty, as determined by the posterior probability calculated by HMMER3. The figure legend explains the different colorings. This view gives a quick and easy visual clue to sequences, or parts of sequences, that align poorly to the HMM.

The seed and full alignments are all available for downloading in a number of different formats, suitable for different alignment-application tools: Selex, FASTA, Pfam or MSF formats. For occasions when users wish to analyze the full-length sequences of members of a domain, they can download the full-length sequences under the "Downloading options" table. This table also allows users to download all alignments in a gzip compressed form.

The "Alignments" tab also gives access to the external site of MyHits,⁷ which provides a collection of tools for handling multiple sequence alignments. For example, the user can refine a seed alignment (by sequence addition or removal, re-alignment or manual editing) and then search databases for remote homologues using HMMER3.

Trees: We generate neighbor-joining phylogenetic trees (with bootstrap values based on 100 replicates shown on tree nodes) for all our alignments, using FastTree⁸ (http://www.microbesonline.org/fasttree/). The user can view the seed, full, NCBI and metagenomic trees by selecting the required alignment at the top of the page and using the "change tree" button. The trees can also be downloaded in Newick format using the links at the bottom of this page.

Curation and Model: This tab contains information about how the HMM was built, including the curated thresholds for the family. The HMM can be downloaded from this page. The source of the seed alignment is also indicated in this section. In the case of the BLIP family (see Figure 1), the seed source is given as Pfam-B_41444 (release 10.0). Many seeds are derived from Pfam-B alignments, particularly when a function has been determined or a structure has been solved for one of the members. In some instances the seed source can be from a literature reference, when it is indicated by a "[1]" (that is, having come from the first reference shown on the "Summary" tab).

Species: This tab shows, by default a "Sunburst" rendition of the species distribution of all the sequences in the full alignment of the family (Figure 13). "Sunburst" uses a radial layout to indicate the species. The sunburst is color-coded according to the taxonomic distribution, thereby allowing rapid assessment of the taxonomic diversity found within the family. It is not interactive. This view has been implemented, because many species trees are very large and display is slow.

The interactive taxonomic tree view is also available in the adjacent tab (Figure 14). The floating tool panel on the right allows the user to toggle between different views of the distribution: The tree can be expanded or collapsed at each of the nodes; those species found in the seed can be highlighted; individual nodes of the tree can be selected and the selected members viewed either as Pfam graphics or as an alignment; selected sequences can be downloaded. Full details on the interactive nature of the tree with explanation of the colors used are given in the legend to Figure 14.

HMM logo: The HMM logos offer a means of visualizing profile HMMs using the LogoMat-M software⁹; logos provide an overview of HMMs in graphical form and help to point up key conserved sequence motifs. As an example we have given the logo for the very short zinc-finger family, zf-C2H2 (Pfam accession PF01530) (Figure 15), where the characteristically conserved cysteines and histidines are clearly shown.

To read a HMM logo, consider the stack of letters of different heights; the height of each stack represents the relative entropy of the distribution of the emission probabilities of the amino acid residues. In essence, the larger the letter the more likely that that amino acid will be present in that respective position in the protein family, and letters are sorted in descending order depending on their relative probabilities. The width of the column also varies to reflect the relative contribution of that position to the overall score; the pink, empty, columns represent the insert states. The colors of the amino acids reflect their respective biological properties, from charged, through polar uncharged, and aliphatic to aromatic, each of these sets being colored through red-brown, blue-purple, orange-yellow to green, respectively.

Interactions: The "Interactions" tab shows the other Pfam domains with which the family of interest interacts. Interaction data are taken from iPfam,¹⁰ a resource that describes physical interactions between Pfam domains that have a representative structure in the Protein Data Bank (PDB). Clicking on the "Interactions" tab for PDZ (Pfam accession PF00595) shows that it interacts with Peptidase_S41 (Pfam accession PF03572), Ras (Pfam accession PF00071) and Trypsin (Pfam accession PF00089). This feature can be useful for investigating biochemical pathways or complexes, and allows consideration of the relationship of the particular domain with other domains.

Structures: Where structures are available for a UniProtKB sequence in the PDB,¹¹ these will be displayed under the "Structure" tabs on the sequence, family or clan pages in the website. They will also each have their own structure page.

For those sequences that have a structure in the PDB, we use the mapping provided by PDBe (http://www.ebi.ac.uk/pdbe/docs/References.html) between UniProtKB and the PDB coordinate systems to map Pfam domains onto three-dimensional protein structures.

The "Structure" tab will show a table of the exact mappings for all structures that map to the Pfam domain of interest. Users can choose one of the following molecular viewer applets: Jmol (http://www.jmol.org), AstexViewer¹² or SPICE,¹³ to view the domain(s). The Jmol viewer and AstexViewer show a representation of the whole PDB structure with the domain of interest marked in green and any other Pfam domains shown in other colors. Any regions that do not map to a Pfam domain are shown in gray, and the images are rotatable and with zoom facilities. In AstexViewer, the protein backbone of the whole PDB structure is drawn as secondary-structure elements, with the domains themselves surrounded by semitransparent molecular surfaces that one can toggle on (Figure 16a) or off (Figure 16b). SPICE is a more complex and powerful viewer that allows the user to browse sequences, structures and their annotations on both the sequence and the structure of a particular protein.

Sequence pages

Each sequence in the primary databases has its own sequence page, which can be accessed in many different ways as indicated above (see Figure 2 for a typical page). The page shows the summary information and a brief description of the sequence from its parent database of UniProtKB, NCBI or, in the case of metagenomic sequences, the source of the data. The precalculated organization of domains and features on the sequence are given with a table indicating the residue ranges for each feature. There is mark-up for active-site residues and metal-binding residues, and the positions of any disulfide bridges are indicated, the residues positions for all of which appear in a pop-up window when mousing over each one. Positions for signal peptide, coiled coil, transmembrane and/or low-complexity regions are also shown, if present. Where there is a corresponding structure in the PDB, this will be displayed on the "Structures" tab. Where a TreeFam¹⁴ phylogenetic tree is available, this is displayed under the "TreeFam" tab.

Clan pages

The clan pages can be accessed in a number of ways. The user can enter a clan accession or ID into the "Jump to" or "View a clan" box on the Pfam home page; use the "Browse" menu at the top of each Pfam page to view a list of clans; or click on the "Clan" tab on a family page. The tabs on the clan page are very similar to some of those on the family page described above, and include "Summary," "Domain organization," "Alignments," "Species," "Interactions" and "Structures" (Figure 4). The clan pages also have a "Relationships" tab, which gives a graphical image of the interconnections between the clan members, determined by HHsearch¹⁵ alignment between the family HMMs (Figure 17). Clicking on any E-value in this graph will take the user to the pairwise HMM logo for that particular relationship, a feature that is particularly useful for comparing the sequence similarity between the two families.

Structure pages

All the data collated on a PDB structure are displayed here together with relevant structure-specific tabs (Figures 5 and 18). The front page (Figure 5) shows the source reference, links to structure-specific external databases, and additional information from the TOPSAN wiki if available (Figure 5). Where a structure has been cited in publications that are available via PubMed Central (http://www.ncbi.nlm.nih.gov/pmc/), these articles can be viewed in marked-up format via BioLit¹⁶ (Figure 18).