Example 1

The C-terminal region (residues 302 – 445) of Q9HAC7 (CaiB/baiF CoA-transferase family protein C7orf10) was a Human region identified as not covered by Pfam, that had >3,000 phmmer [1] hits in UniProtKB. Just over 40% of this protein was covered by PF02515 (CoA-transferase family III) in Pfam 27.0 (Figure 1A). There are several structures available for bacterial members of this family, which confirm that the domain boundaries on this family could be widened. I successfully extended the family at both the C-terminus and the N-terminus, and Pfam now covers 80% of Q9HAC7 (Figure 1B). The extended family will be released in Pfam 28.0.

Example 2

Some cases are more complex and require improvements to be made to several different families. A region of P46063 (ATP-dependent DNA helicase Q1, RECQ1) covering residues 410 to 487 was identified as having no Pfam-A domain and had >550 phmmer hits in UniProtKB. This protein is covered by three Pfam-A domains and a Pfam-B family in Pfam 27.0 (Figure 2A). Three-dimensional structures are available for this protein and for the Escherichia coli RecQ protein [2-3], and show RecQ to contain four distinct domains. The N-terminus of the protein is a DEAD/DEAH box helicase domain (PF00270), followed by a helicase conserved C-terminal domain (PF00271), then a zinc-binding domain, which was absent from Pfam 27.0, and, at the C-terminus, an RQC domain (PF09382). The Pfam alignment for the helicase conserved C-terminal domain was too short, so I extended this at the N-terminus to cover the region matched by the Pfam-B family. Extensions to large families such as this one (over 89,000 members in Pfam 27.0) help to increase the Pfam residue coverage of UniProtKB, something we are always striving to do. A new family, PF16124, was built for the zinc-binding domain, filling the gap between the helicase conserved C-terminal domain and the RQC domain, and the RQC domain was extended by a small amount at both ends. Figure 2B shows the new Pfam architecture of this protein, which is a far better reflection of it’s structure. The changes to PF00271 and PF09382, and the new family PF16124 will be available in Pfam 28.0.

Example 3

In some cases there are no appropriate changes to be made to existing families or new families to build, an example of this is P01138 (Beta-nerve growth factor). The N-terminal region of this protein is not covered by Pfam and has over 600 hits to UniProtKB. However, closer examination revealed that the N-terminus of this protein consists of a signal peptide and a propeptide and it would therefore not be appropriate to include these regions in Pfam.

Our study on the human proteome [4] has so far proved to be useful for selecting interesting regions to target for family building. We will continue to work on the long list of conserved Human regions over the next few months.

Posted by Ruth

References

[1] http://hmmer.janelia.org/

[3] Pike, A.C., et al. (2009) Structure of the human RECQ1 helicase reveals a putative strand-separation pin, Proc Natl Acad Sci U S A, 106, 1039-1044.

[4] Mistry, J., et al. (2013) The challenge of increasing Pfam coverage of the human proteome, Database (Oxford), 2013, bat023

Within the human proteome, we identified almost 15,000 clusters of sequence regions that are not in Pfam-A or Pfam-B families, not predicted to be signal peptide regions, and that are at least 50 residues long. This accounts for close to 40% of all human residues (Figure 1, click on the figures to see a higher resolution version of them).

Figure 1: Proportion of human residues covered by Pfam, in signal peptide regions predicted by Phobius [3], and in small regions unlikely to harbour Pfam domains. Click on the figure to see a higher resolution image.

Figure 2: Distribution of the mean number of significant hits found in each cluster after a phmmer search against UniProtKB by each cluster member. Each cluster (~15,000 in total) represents a set of homologous HRs that do not contain a Pfam-A domain. Click on the figure to see a higher resolution image.

This histogram has two important features:

1. most HRs have few hits in UniProtKB (we will comment more specifically on these regions in a future blog post), and
2. there is a long tail of regions with many hits in UniProtKB.

Want to guess which regions Pfam will target first? Of course we want to go for regions that might give us maximum impact, so we will start by trying to incorporate into Pfam the HRs that have the highest number of hits in UniProtKB. In fact, add to this the plot in Figure 3 below and you have our current strategy for adding conserved HRs to Pfam.

Figure 3: Distribution of the mean proportion of hits that have overlap to Pfam-A, for clusters with a mean phmmer output of >500 hits in UniProtKB. Click on the figure to see a higer resolution image.

Figure 3 was drawn by calculating, for each cluster member, the percentage of its phmmer hits that matched an existing Pfam-A family, and averaging these values over all members in the cluster. We show only clusters that have >500 phmmer hits in UniProtKB. We see that a large number of regions have a high percentage of such ‘overlaps’ with existing Pfam-A families. These HRs are likely to be regions that are on the periphery of existing families (outliers) and that have been narrowly missed by our profile-HMMs. Alternatively, they could be problem regions that generate overlaps with many different clans, such as is sometimes observed for coiled-coil regions (see our recent paper [5] on the subject). You will also notice, however, that a large proportion of HRs feature few or no overlaps to Pfam-A. These are the ones that we are currently trying to add to Pfam. Ruth Eberhardt, one of our curators, will give some examples of HRs that she has recently added to Pfam-A in another post shortly.

It’s still very early stages in this analysis and we have looked at only around 30 cases. About two thirds resulted in extensions to existing families, and 6 led to new families being built. In a number of instances, such as the ones that will be discussed in our next post, structural information was key to improving family annotation. In many other cases, the only information available was sequence conservation, and based on this our curators had to make a decision as to whether or not to extend an existing family or to create a new one.

Overall, it looks as if the analysis of conserved human regions has given us great potential not only for increasing Pfam coverage of human, but also for improving the boundaries of many of our families. We can imagine that in the future, a similar strategy could be applied to other sets of sequences such as those, for example, in the proteomes of model organisms or pathogens.

Posted by Jaina and Marco

References

[1] Mistry, J., et al. (2013) The challenge of increasing Pfam coverage of the human proteome, Database (Oxford), 2013, bat023.

[2] (2013) Update on activities at the Universal Protein Resource (UniProt) in 2013, Nucleic Acids Res, 41, D43-47.

[3] Kall, L. et al. (2004) A combined transmembrane topology and signal peptide prediction method. J Mol Biol, 338, 1027-1036.

[4] http://hmmer.janelia.org/

[5] Mistry, J., et al. (2013) Challenges in homology search: HMMER3 and convergent evolution of coiled-coil regions, Nucleic Acids Res.

TreeFam 9 now has 109 species (vs. 79 in TreeFam 8) and is based on data from Ensembl v69, Ensembl Genomes v16, Wormbase and JGI.

This release marks an important step for TreeFam as it is the first release build since TreeFam has been resurrected.
Here is a list of the most important changes in TreeFam 9:

• New website layout (adopting the Pfam/Rfam/Dfam layout)
• Infrastructure move of web servers and databases to the EBI
• Sequence search against the library of TreeFam family profiles
• new tree visualisations in pure javascript using D3, e.g. see the BRCA2 gene tree here.
• Pairwise homology download

We hope you find all the information you are looking for. If you don’t, please let us know so that we can include the information you want. The old website will remain online here.

If you have questions, suggestions or find bugs, don’t hesitate to contact us through our new forum here.

Happy treefamming,

the TreeFam team
(Fabian, Mateus)

So what has changed?  To the user, hopefully, not a great deal has changed! Nevertheless, there has been a considerable amount of reorganization of the database production pipeline. The most notable loss of information is that we are no longer providing neighbour-joining trees for the Pfam full alignments.  If you want or care about this type data,  it will now be up to you to calculate it – however, if this data was important to you it should be recalculated using a more precise method anyway.  On the flip side, there are many new features that have been integrated into Pfam 27.0.  Below is a brief list of the new developments that are now available:

• Real time searches of DNA  sequences for matches to Pfam models
• Use of Representative proteomes (1) sequence sets used to provide redundant views of the Pfam-A full alignments
• Addition of disorder predictions to the repertoire of sequence feature annotations
• AntiFam (2) has been applied to the underlying sequence database to remove sequences believed to be spurious translations
• Selectable sunbursts in the Pfam-A ‘species’ distribution tab, allowing the generation of alignments or visualisation of sequences from a user defined taxonomic range.
• New, faster keyword search using Apache Lucy

Blog posts describing these news developments in more detail will follow in the coming weeks.  In addition to the changes listed above, there have been many improvements to existing families, be it improving domain boundaries, expanding members or the generation of Wikipedia entries.  Many of the  new entries in Pfam have been built with the purpose of improving Human coverage.

Enjoy the release!

Posted by Rob

References

1: Chen C, Natale DA, Finn RD, Huang H, Zhang J, Wu CH, Mazumder R. Representative proteomes: a stable, scalable and unbiased proteome set for sequence analysis and functional annotation. PLoS One. 2011 6(4):e18910

2: Eberhardt RY, Haft DH, Punta M, Martin M, O’Donovan C, Bateman A. AntiFam: a tool to help identify spurious ORFs in protein annotation. Database 2012:bas003.

In other exciting news, two members of the Dfam consortium, Arian Smit and Robert Hubley (Institute for Systems Biology, Seattle), just released RepeatMasker 4.0. This is a major update that, among other important improvements, adds support for searching with Dfam and nhmmer. Go get yourself a copy at http://www.repeatmasker.org/

Posted by Travis

TreeFam 9 will have 109 species, that is a 37% increase over TreeFam 8. Most of the species come from EnsEMBL (v.69) and EnsEMBL genomes (v.16) with a few ones coming from JGI.

Besides that – and probably most important for the user – will be our new web site. Based on the success of other Xfam-databases like Pfam [5], Rfam [6] and -most recently- Dfam [7], we decided to give the TreeFam website a face lift by adapting it to the Xfam look&feel.

So, there are great things to come and soon we will have our next blog post.
The next TreeFam blog post will then be about TreeFam 9!

The R-chie diagrams are created using the R4RNA R package from Irmtraud Meyer’s group; be sure to check out the R-chie paper, as well as their own gallery of Rfam structures.

New Hit Results

The underlying database and set of entries have not changed from Dfam 1.0, but the seed alignments have been tweaked to add in some additional hits from chromosome Y and the sequence matches (and match boundaries) have changed slightly as a result of using a new version of nhmmer (snapshot here) that better handles regions of low complexity. This allowed us to mask fewer low-complexity regions in models and decrease gathering thresholds (increasing sensitivity) while still slightly reducing false positive rates.

Relationships tab

In many cases, a single locus of DNA sequence might be matched by multiple Dfam models, due to similarity between models. For example, there are 37 Alu subfamily models, all representing minor variants of the prototypical Alu, so any Alu instance will likely be found by nearly all Alu models. Many models have a more complicated relationship, as is the case with Ricksha (which long ago picked up the 3′ end of an ERVL, including its LTR, MLTB2), and SVA (which carries copies of both a portion of a HERVK LTR and two Alus in reverse orientation). To indicate these sorts of relationships between models, we have created a Relationships tab for each model, where we graphically represent related models, and provide a dot plot for each relationship that depicts the alignment between the related models.

Redundant profile hit resolution

We call these cases of a single locus matching multiple profile HMMs, “Redundant Profile Hits” (RPHs). When presenting hit counts and hit distributions both over the model (Coverage Plot on the Model tab) and over the genome (Hit tab), Dfam 1.0 only showed the total number of hits, so that an RPH would impact multiple model tallies. We now resolve RPHs, so that each locus is associated with only the model that best explains it (though in some cases, the resolution choice is unclear, and a couple of models might be assigned to the locus).

And finally…..

We have already started to receive positive feedback from some early adopters of the databases.  We really appreciate this and suggestions for new features, so if you have any suggestions or would like to contribute models to the database, please get in contact.  In the next release of Dfam, we aim to continue making improvements to the existing set of entries, before we pause and determine the next major milestone.

Posted by Travis and Rob

 New features

AntiFam has several new features, which we hope our users will find useful. In release 2.0 we introduced a DO line, this is the line in the AntiFam.seed file which begins “#=GF DO”. This line gives a recommended course of action to take when a predicted protein sequence is recognised by an AntiFam HMM. In most cases this line recommends that the predicted protein is deleted.

In release 3.0 we have introduced a TX line, the line in the AntiFam.seed file which begins “#=GF TX”. This is used to indicate which superkingdoms predicted proteins recognised by AntiFam HMMs have been found in. We have also produced superkingdom-specific sets of HMMs for this release.  One AntiFam HMM may identify spurious proteins arising from multiple superkingdoms, and therefore there is some overlap between these sets.  There are 18 entries in the Eukaryota set, 47 in the Bacteria set, 4 in the Archaea set, 4 in the Virus set and 18 in the unidentified set; the latter includes proteins from unclassified organisms, such as those from metagenomics studies.

Availability

AntiFam 3.0 is available to download from our ftp site on ftp://ftp.sanger.ac.uk/pub/databases/Pfam/AntiFam/

AntiFam is made freely available under the Creative commons Zero (CC0) licence http://creativecommons.org/publicdomain/zero/1.0/

We welcome any suggestions for new families, and can be contacted by email ruthe@ebi.ac.uk

Posted by Ruth