Bioinformatics Advance Access originally published online on July 19, 2005
Bioinformatics 2005 21(18):3691-3693; doi:10.1093/bioinformatics/bti589
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
PACdb: PolyA Cleavage Site and 3'-UTR Database
1The Jackson Laboratory Bar Harbor, ME 04609, USA
2Bioinformatics Program, Boston University Boston,MA 02215, USA
3Functional Genomics Program, University of Maine Orono, ME 04469, USA
*To whom correspondence should be addressed.
 |
Abstract |
|---|
 TOP Abstract 1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
Summary: The PolyA Cleavage Site and 3'-UTR Database (PACdb) is a web-accessible database that catalogs putative 3'-processing sites and 3'-UTR sequences for multiple organisms. Sites have been identified primarily via expressed sequence tag-genome alignments, enabling delineation of both the specificities and heterogeneity of 3'-processing events.
Availability: By web browser or CGI: PACdb: http://harlequin.jax.org/pacdb/; AtPACdb: http://harlequin.jax.org/atpacdb/
Contact: joel.graber{at}jax.org
Supplementary information: Available online at http://harlequin.jax.org/pacdb/supplemental.php
|
1 INTRODUCTION |
|---|
|
TOPAbstract1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
Most eukaryotic mRNA molecules terminate in a polyadenylate [poly(A)] tail that can affect mRNA localization, stability and translational efficiency (Zhao et al., 1999). Many genes have been shown to utilize alternative 3'-processing sites depending on the tissue type and/or environmental conditions at the time of processing (Edwalds-Gilbert et al., 1997; Sparks and Dieckmann, 1998; Gautheret et al., 1998; Beaudoing et al., 2000; Kan et al., 2001). Such alternate 3'-processing can result in altered regulation of the gene, including near complete silencing of the gene in extreme cases (van Hoof et al., 2002).
The location and efficiency of 3' end formation is controlled by a complex interaction between cis- and trans-acting factors. The proteins involved are highly conserved across a broad range of organisms (Keller and Minvielle-Sebastia, 1997). Although the sequence content of cis-acting elements is often conserved, the positioning can vary widely, especially in comparisons of animal and plant (or yeast) sequences, in which a large-scale rearrangement is apparent. The canonical hexamer A(A/U)UAAA, once thought to be nearly ubiquitous, has been shown to be much more variable, depending on organism and even tissue or environmental conditions (MacDonald and Redondo, 2002).
We present here the PolyA cleavage site and 3'-UTR Database (PACdb), in which we have used transcript–genome alignments, primarily expressed sequence tags (ESTs), to identify and characterize putative 3'-processing sites for human, mouse, rat, dog, chicken, zebrafish, fugu, fruitfly (Drosophila melanogaster), mosquito, nematode (Caenorhabditis elegans), Arabidopsis thaliana, rice (Japonica) and baker's yeast. See the Supplementary tables for statistics on each organism as of June 24, 2005 and visit the website for live information. Our system consists of three linked databases holding EST data, alignment coordinates, and putative 3'-processing sites. We characterize all putative 3'-processing sites for each gene and assign a confidence level to each site. The information is freely available via our web server.
Several existing resources identify putative 3'-processing sites and/or 3'-UTR sequences. PolyA_DB (Zhang et al., 2005) uses methodology similar to ours; however, it is currently limited in scope to mouse and human and more restrictive in 3'-processing site identification. UTRdb (Mignone et al., 2005) is generated via careful parsing of EMBL/Genbank records, but it focuses on UTR sequences, with no specific focus on 3'-processing sites.
|
2 METHODS AND IMPLEMENTATION |
|---|
|
TOPAbstract1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
We developed an automated process to determine putative 3'-processing sites, with four general steps: EST preprocessing, EST-genome alignment, gene mapping and site characterization. The data are fed into three linked databases holding EST data, alignment coordinates and putative 3'-processing sites. With slight modification, the same process is used for cDNA. Specific details of our methodology are available in the Supplementary information; in this paper, we highlight the aspects of our approaches that differ from or extend previous approaches (Gautheret et al., 1998; Beaudoing et al., 2000; Kan et al., 2001; Zhang et al., 2005; Yan and Marr, 2005).
Putative 3'-processing sites are characterized according to both genomic context and supporting cDNA/EST sequences. The genomic context is assessed for count of supporting sequences and flanking A-rich sequence or restriction enzyme cleavage sites (exact and near-matches), either of which indicate potential false 3' ends. For each supporting cDNA/EST, we track the uniqueness of genomic alignment and evidence of a polyA tail. We combine the genomic and sequence properties to assign a confidence level (very high, high, medium, low, very low, summarized in Supplementary Table 3) for each putative 3'-processing site.
Confidence level assignment is useful in determining whether a putative 3'-processing site is actively used. However, lack of assignment as either ‘very high’ or ‘high’ confidence does not necessarily negate a 3'-processing site. For instance, rarely expressed genes or rarely selected 3'-processing sites would correspondingly be rarely observed in the EST data, decreasing the confidence assignment. Also, in the plant A.thaliana, nearly all EST tails were clipped before being dumped into public repositories (see ‘PolyA Tail Evidence’ in Supplementary Table 2) and therefore, cannot attain ‘very high’ confidence since that data were not reported. Because of these and other similar examples, PACdb stores all putative 3'-processing sites, but provides the properties and confidence levels to help users intelligently filter false sites (see Supplementary Table 4 for results of confidence level assignment across organisms in PACdb).
PACdb is accessible through a web interface that includes simple and advanced query forms. The advanced search form allows restriction by multiple features including, but not limited to, organism, chromosome/contig, number of cleavage sites per gene, Gene Ontology annotations (Gene Ontology Consortium, 2000), gene name/description, tissue information and more. Search results include genes, 3'-processing site genomic flanking sequence, 3'-UTR sequence, EST information and 3'-processing site details, and can be retrieved in a variety of formats, including tab delimited text, HTML Table and FASTA (when outputting sequence). Forms are also provided for homology-based searches using either protein or nucleic acid sequences, and targeting either annotated genes or genomic sequence. Genomic searches are particularly useful for identifying 3'-processing sites for either unannotated or non-protein coding genes.
We have developed a simple web API to allow URL-based retrieval that will allow other web-based databases to easily connect to PACdb. A full description of the interface can be found on the PACdb website. Owing to specific interest, we have implemented a parallel HTML interface for accessing just A.thaliana information. This information can also be accessed using the PACdb interface.
|
3 RESULTS AND DISCUSSION |
|---|
|
TOPAbstract1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
PACdb displays the 3'-processing sites for a specific gene in a graphic that includes gene structure, aligned ESTs and putative 3'-processing sites (Fig. 1A). The image may be customized in several ways, including restricting confidence level, toggling labels for genomic A-rich regions or restriction enzyme sites adjacent to a 3'-processing site and library-specific coloring of the ESTs. Such library-specific labeling of ESTs offers the possibility of highlighting 3'-processing sites specific to tissue or developmental stage.
|
As seen in Figure 1A, 3'-processing sites identified via EST-genome alignments commonly show heterogeneous cleavage (Kan et al., 2001; Zhang et al., 2005; Yan and Marr, 2005), in which multiple close neighboring sites occur without apparent regulatory implications. Clustering adjacent putative 3'-processing sites reduces both the number of singleton ESTs and putative 3'-processing sites, depending on window size; however, clustering can lead to an artificial grouping of separate cleavage sites if too large of a clustering window size is arbitrarily assigned. The empirically observed distribution of separation between 3'-processing sites within a gene is illustrated in Figure 1B for selected organisms in PACdb. All organisms investigated to date have disjoint exponential distributions with variable coefficients that depend on the site-to-site separation. The transition between realms of distinct coefficient provides a reasonable threshold for clustering neighboring sites, yet as Figure 1B shows, such thresholds appear to be organism-specific (e.g.
20–25 nt for mammals, but 10–15 nt for fish or plants). In contrast to previous similar studies, we explicitly do not eliminate potentially false 3'-processing sites, but rather collect a broad range of evidence for distinguishing false positives for the confidence level assessment. PolyA_DB (Zhang et al., 2005) has the requirement that all putative 3'-processing sites have evidence of a polyA tail. However, because a large number of ESTs have their tails clipped before being made publicly available (especially for A.thaliana), PACdb catalogs if an EST has evidence of a tail, but does not require it. This information is then used in determining the ‘confidence level’ of the putative 3'-processing site. Web interface users can specify which confidence level they prefer, and thus can decide for themselves whether to require evidence of a tail.
One specific advantage to the inclusion of (but assignment of lower confidence to) potentially false sites in PACdb is the possibility for comparison with gene-expression measurements, such as SAGE (Velculescu et al., 1995) and MPSS (Brenner et al., 2000). Such measurements are generated using procedures similar to EST generation and are therefore subject to the same systematic sources of false 3'-end generation.
Another specific difference in our approach is that we do not explicitly require the assignment of putative 3'-processing sites to an annotated gene. In contrast, other groups (Zhang et al., 2005; Yan and Marr, 2005) require that a cDNA/EST–genome alignment overlap with a RefSeq mRNA–genome alignment in order to be mapped to a gene, a constraint that could result in the removal of legitimate distal cleavage sites. In the process of mapping a cDNA/EST to a gene, we use the cDNA/EST–genome alignment and the likely orientation to map to the nearest gene within an organism-specific threshold. Thresholds are determined by first mapping sites to the next correctly oriented upstream gene with no restriction on distance. As shown in Figure 1C for mouse and A.thaliana, the distribution of the distance from putative 3'-processing site to the next gene is typically bimodal. The second (longer) mode is apparently owing to missing gene annotations, as evidenced by comparison with the separation between consecutive, commonly oriented, genes (Fig. 1C). We use a least-squares approach to fit the empirical distribution of 3'-processing site to stop codon distance as a mixture of two lognormal distributions (data not shown). Finally, the threshold for maximum allowed distance for gene assignment of a 3'-processing site is set at the distance where the contribution of the two lognormal components to the total is equal.
If an aligned EST cannot be mapped to a gene within the threshold distance, the EST is stored in PACdb as a putative 3'-processing site but marked as unmapped. For well-annotated organisms, these unmapped sites are good candidates to be polyadenylated, non-protein coding RNA (ncRNA) genes.
PACdb currently uses gene annotations and IDs from multiple sources including Ensembl (Hubbard et al., 2002) and organism-specific databases, such as SGD (Dwight et al., 2004) and TAIR (Rhee et al., 2003). Currently, there is only one gene annotation set per organism, but future plans include the use of multiple gene annotation sources for each organism where possible.
A final distinction of PACdb is the inclusion of data from a diverse set of organisms including four mammals, two fish, two insects, one bird, one nematode, two plants and one fungus. The organism diversity in PACdb allows for more cross-genomic comparisons and could aid in finding conserved 3'-UTR cis-elements and also for comparing whole-organism polyadenylation patterns.
|
4 CONCLUSION |
|---|
|
TOPAbstract1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
PACdb is a database that currently stores putative 3'-processing sites for human, mouse, rat, dog, chicken, zebrafish, fugu, fruitfly (D.melanogaster), mosquito, nematode (C.elegans), A.thaliana, rice (Japonica) and baker's yeast. Data updates will occur regularly to incorporate new EST/cDNA sequences, and genomic updates will occur as new genome drafts become available. Future work will involve importing additional organisms, broadening gene annotation data and adding new capabilities and tools to the web interface. As we continue to collect the 3'-processing information for various organisms, we will develop predictive models to aid in prediction of 3'-processing sites and also augment existing gene prediction algorithms. Finally, we also plan to look at ways of connecting to existing, related resources to provide the user with additional evidence of the usage and regulatory implications of 3'-processing sites.
|
Acknowledgments |
|---|
The authors thank Ray Lauer for the preliminary work and Carol Bult and Martin Ringwald for the critical reviews of the manuscript. This research was partially supported by NSF research contract DBI-0331497, NIH/NCRR INBRE Maine contract 2 P20 RR16463-04, NSF IGERT award number DGE-9870710, and NSF IGERT grant 0221625. Funding to pay the Open Access publication charges for this article was provided by The Jackson Laboratory.
Conflict of Interest: none declared.
|
Footnotes |
|---|
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors. 
Received on March 25, 2005; revised on June 25, 2005; accepted on July 14, 2005
|
REFERENCES |
|---|
|
TOPAbstract1 INTRODUCTION2 METHODS AND IMPLEMENTATION3 RESULTS AND DISCUSSION4 CONCLUSIONREFERENCES |
|---|
Beaudoing, E., et al. (2000) Patterns of variant polyadenylation signal usage in human genes. Genome Res, 10, 1001–1010
Brenner, S., et al. (2000) Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays. Nat. Biotechnol, 18, 630–634[CrossRef][ISI][Medline].
Dwight, S.S., et al. (2004) Saccharomyces genome database: underlying principles and organisation. Brief. Bioinformatics, 5, 9–22
Edwalds-Gilbert, G., et al. (1997) Alternative poly(A) site selection in complex transcription units: means to an end? Nucleic Acids Res., 25, 2547–2561
Gautheret, D., et al. (1998) Alternate polyadenylation in human mRNAs: a large-scale analysis by EST clustering. Genome Res, 8, 524–30
Gene Ontology Consortium. (2000) Gene Ontology: tool for the unification of biology. Nat. Genet, 25, 25–29[CrossRef][ISI][Medline].
Hubbard, T., et al. (2002) The Ensembl genome database project. Nucleic Acids Res., 30, 38–41
Keller, W. and Minvielle-Sebastia, L. (1997) A comparison of mammalian and yeast pre-mRNA 3'-end processing. Curr. Opin. Cell Biol, 9, 329–336[CrossRef][ISI][Medline].
Kan, Z., et al. (2001) Gene structure prediction and alternative splicing analysis using genomically aligned ESTs. Genome Res, 11, 889–900
MacDonald, C.C. and Redondo, J.L. (2002) Reexamining the polyadenylation signal: were we wrong about AAUAAA? Mol. Cell Endocrinol., 190, 1–8[CrossRef][ISI][Medline].
Mignone, F., et al. (2005) UTRdb and UTRsite: a collection of sequences and regulatory motifs of the untranslated regions of eukaryotic mRNAs. Nucleic Acids Res, 33, 141–146[CrossRef].
Rhee, S.Y., et al. (2003) The Arabidopsis Information Resource (TAIR): a model organism database providing a centralized, curated gateway to Arabidopsis biology, research materials and community. Nucleic Acids Res, 31, 224
Sparks, K.A. and Dieckmann, C.L. (1998) Regulation of poly(A) site choice of several yeast mRNAs. Nucleic Acids Res, 26, 4676–4687
van Hoof, A., et al. (2002) Exosome-mediated recognition and degradation of mRNAs lacking a termination codon. Science, 295, 2262–2264
Velculescu, V.E., et al. (1995) Serial analysis of gene expression. Science, 270, 484–487
Yan, J. and Marr, T.G. (2005) Computational analysis of 3'-ends of ESTs shows four classes of alternative polyadenylation in human, mouse, and rat. Genome Res, 15, 369–375
Zhang, H., et al. (2005) PolyA_DB: a database for mammalian mRNA polyadenylation. Nucleic Acids Res, 33, D116–D120
Zhao, J., et al. (1999) Formation of mRNA 3' ends in eukaryotes: mechanism, regulation, and interrelationships with other steps in mRNA synthesis. Microbiol. Mol. Biol. Rev, 63, 405–445
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Mangone, P. MacMenamin, C. Zegar, F. Piano, and K. C. Gunsalus UTRome.org: a platform for 3'UTR biology in C. elegans Nucleic Acids Res., January 11, 2008; 36(suppl_1): D57 - D62. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. Graber, J. Salisbury, L. N. Hutchins, and T. Blumenthal C. elegans sequences that control trans-splicing and operon pre-mRNA processing RNA, September 1, 2007; 13(9): 1409 - 1426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Liu, J. M. Brockman, B. Dass, L. N. Hutchins, P. Singh, J. R. McCarrey, C. C. MacDonald, and J. H. Graber Systematic variation in mRNA 3'-processing signals during mouse spermatogenesis Nucleic Acids Res., January 12, 2007; 35(1): 234 - 246. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Y. Lee, I. Yeh, J. Y. Park, and B. Tian PolyA_DB 2: mRNA polyadenylation sites in vertebrate genes Nucleic Acids Res., January 12, 2007; 35(suppl_1): D165 - D168. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Vinther, M. M. Hedegaard, P. P. Gardner, J. S. Andersen, and P. Arctander Identification of miRNA targets with stable isotope labeling by amino acids in cell culture Nucleic Acids Res., September 11, 2006; 34(16): e107 - e107. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||