PepBank - a database of peptides based on sequence text mining and public peptide data sources

1471-2105-8-280 1471-2105 Database PepBank - a database of peptides based on sequence text mining and public peptide data sources Shtatland Timur tshtatland@partners.org Guettler Daniel dguettler@partners.org Kossodo Misha dalten@gmail.com Pivovarov Misha mpivovarov@partners.org Weissleder Ralph weissleder@helix.mgh.harvard.edu

Center for Molecular Imaging Research, Massachusetts General Hospital, Harvard Medical School, Bldg. 149, 13^thStreet, Room 5406, Charlestown, MA 02129, USA

Northern Essex Community College, 100 Elliott Street, Haverhill, MA 01830, USA

BMC Bioinformatics 1471-2105 2007 8 1 280 http://www.biomedcentral.com/1471-2105/8/280 17678535 10.1186/1471-2105-8-280 16 3 2007 01 8 2007 01 8 2007 2007 Shtatland et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background

Peptides are important molecules with diverse biological functions and biomedical uses. To date, there does not exist a single, searchable archive for peptide sequences or associated biological data. Rather, peptide sequences still have to be mined from abstracts and full-length articles, and/or obtained from the fragmented public sources.

Description

We have constructed a new database (PepBank), which at the time of writing contains a total of 19,792 individual peptide entries. The database has a web-based user interface with a simple, Google-like search function, advanced text search, and BLAST and Smith-Waterman search capabilities. The major source of peptide sequence data comes from text mining of MEDLINE abstracts. Another component of the database is the peptide sequence data from public sources (ASPD and UniProt). An additional, smaller part of the database is manually curated from sets of full text articles and text mining results. We show the utility of the database in different examples of affinity ligand discovery.

Conclusion

We have created and maintain a database of peptide sequences. The database has biological and medical applications, for example, to predict the binding partners of biologically interesting peptides, to develop peptide based therapeutic or diagnostic agents, or to predict molecular targets or binding specificities of peptides resulting from phage display selection. The database is freely available on http://pepbank.mgh.harvard.edu/, and the text mining source code (Peptide::Pubmed) is freely available above as well as on CPAN (http://www.cpan.org/).

Background

Peptides have emerged as important affinity ligands for diagnostic and therapeutic medical uses as well as materials for a host of applications in biotechnology. While many excellent databases exist that provide protein sequence data 123, protein interaction data 456789, and peptide data 10111213, a substantial fraction of literature data remains untapped. Unfortunately, the wealth of the peptide sequences in these sources is often difficult to access by modern methods of sequence similarity searching, because peptide sequences are not extracted in a suitable format. We therefore sought to address this issue by developing a combination of automatically mining MEDLINE abstracts for peptide sequences, combining the existing bioinformatics sources, and manually curating the full text articles and MEDLINE text mining results. The data, available through a web-based interface for simple and more advanced text search and BLAST and Smith-Waterman sequence similarity search, proved useful in our own work. Examination of initial data yielded some surprises as well, providing an incentive for us to make further improvements to the database. We hope that the peptide database, the associated tools, and the text mining algorithm will be useful to the larger biomedical community.

Peptides are defined by International Union of Pure and Applied Chemistry and International Union of Biochemistry and Molecular Biology (IUPAC-IUB) as compounds "produced by amide formation between a carboxyl group of one amino acid and an amino group of another" 14. In this paper, we use the term "peptides" as a common synonym for oligopeptides, which are defined as having "fewer than about 10–20 residues"14. We thus currently use an IUPAC-IUB length cut-off of 20 amino acid residues or less. Many of the peptides used as pharmaceutical and diagnostic agents fall within this cut-off.

Naturally occurring peptides function as hormones, transmitters, and modulators of numerous biological processes 15. Both naturally occurring and synthetic peptides are used in therapeutic applications 15, for example somatostatin analogs in tumor radiotherapy 1617 and oxytocin to induce labor 18. Examples of diagnostic uses include membrane-translocating agents 19, receptor targeting agents 20, and enzyme substrates 21. Driven by the great interest in the diverse applications of peptides, the new peptidomics field is rapidly emerging 22. The functions of peptides, including their interacting partners, are determined by their sequence and similar to longer proteins, can be predicted based on sequence similarity.

Prior knowledge can be used to predict or shorten the list of possible binding partners of a given peptide of interest, provided a peptide shares significant sequence similarity with other peptides or proteins whose binding partners are known 2023. One can also use a sequence similarity search to remove peptides with similarity to other peptides with known, undesirable properties such as non-specific binding 24 or toxicity. Computational predictions are relatively fast and inexpensive, but require a peptide sequence database with links to peptide data, for use with sequence similarity search methods such as basic local alignment search tool (BLAST) 2526 or Smith-Waterman search 2728. The non-sequence (text) data in such a peptide database can be queried with text search tools for biological, therapeutic or diagnostic applications, for example to find peptides that are enzyme inhibitors and whose sequences are available.

We searched through the existing bioinformatics sources, and found no single source that fully suited our needs. With the exception of the Receptor Ligand Contacts (RELIC) database and web-server 10 and Artificially Selected Proteins/Peptides Database (ASPD) 11, most large protein sequence and interaction databases that allow both sequence similarity and text annotation searches have two major drawbacks. First, most of their sequences are of biological origin, while many phage display 2930 or combinatorial screens yield non-biological sequence hits. There is no large repository of chemically generated unnatural sequences, similar to what PubChem 2 or ChemBank 31 are for compounds. Second, there exists less data on short peptides than on longer proteins, and usually no facile way to restrict the search to short sequences only. This is important because performing an unrestricted sequence similarity search often results in a large proportion of false positives due to hits to proteins in which the peptide sequence is buried and not accessible for binding, or is in a conformation different from that in a shorter peptide. The same sequence may have different binding properties when displayed on a phage versus when presented as part of the native protein 32. Sequence similarity based predictions are further hampered for conformationally constrained peptides, designed specifically to have properties different from the same sequence in linear form 33. ASPD 11 and RELIC 10 databases do not have these drawbacks, are well curated, but are relatively small compared with the large amount of sequence data in the MEDLINE abstracts. For example, the ASPD database has 1,717 entries of 20 amino acid or shorter sequences. RELIC (a server with many useful peptide sequence analysis tools) has 3,632 peptide sequences that result from phage display selections, but only 7 distinct targets to which they bind. Other peptide databases have different purposes and are more specialized by design, for example antimicrobial (the Antimicrobial Peptide Database (APD) 13, and others 123435), phosphorylation sites (Scansite 36), or major histocompatibility complex related (SYFPEITHI 37, EPIMHC 38, and others 39404142).

In order to create a database suitable for the identification of affinity ligands, we developed text mining methods to extract peptide sequences from MEDLINE abstracts and compiled them in a single, easily searchable database. While far from complete, the database is a useful publicly available source of peptide sequences and the associated data. Below we show how the database was constructed, how it functions, and how it can be used to identify targeting ligands.

Construction and content

Database model and overview

The database model (Figure 1) was adopted from the Proteomics Standards Initiative Molecular Interactions (PSI MI) model for storage of biological interactions 43 and was extended to facilitate secure access to curate entries. Each entry is associated with a "peptide sequence", an "interactor", an "experiment" and a "group". The group serves to assign user permissions for curating entries. Separate tables, which are not shown for clarity, define controlled vocabularies. These were adopted where possible from the existing ontologies. Organism vocabulary used for peptides, interactors and interactions was adopted from the National Center for Biotechnology Information (NCBI) Taxonomy 2. The detection method vocabulary, utilized for experiments, was adopted from PSI MI ontology using the descendants of the term MI:0001, "interaction detection method".

Figure 1

Database core model

Database core model. MySQL tables are shown as rectangles. Mandatory attributes are in bold, optional are in italics. Relationships are shown as lines, with the arrows pointing from the primary to the foreign keys, and multiplicities as shown.

The application is using the open source Ruby on Rails framework 44 with a MySQL database 45 in the backend. The BLAST search 2526 was implemented using the NCBI binaries 46. The Smith-Waterman search was implemented using the SSEARCH program from the FASTA3 distribution 272847. The databases for sequence similarity searches included, in addition to sequences, the motifs, with any variable positions replaced with X for simplicity (for example, motif 'P(P/S)GH(Y/F)K' was used as 'PXGHXK').

The database was constructed from the following sources (with the current number of entries in parentheses): text mining of MEDLINE abstracts (13,596 entries), manual curation of full text PDF articles (859), and other public sources: ASPD (1,717) and UniProt (3,620), as described in the sections below. The total number of entries is currently 19,792. A small fraction of the peptide sequences resulting from MEDLINE abstract text mining were manually curated: 1,773 entries were validated as correct peptide sequences, and 170 of those were more fully annotated with additional interaction data present in the abstract. The database continues to grow as the new data are added to the sources such as MEDLINE and UniProt.

MEDLINE abstract text mining

In order to identify abstracts with peptide sequences, the entire MEDLINE database with its 15 million records was downloaded from the National Library of Medicine (NLM) ftp site 48. The text mining code was written in Perl, a language selected due to its text processing capabilities, and widely used in many important biomedical literature text mining applications 495051. Data were processed in 3 steps. First, each abstract was assigned a score based on how likely it was to contain a peptide sequence anywhere within the text. Second, each individual word was assigned a score based on how likely it was to contain a peptide sequence. For each word, a combined score was then computed based on both the word score and the abstract score. Thus, in total, we used three types of scores (abstract, word, and combined). Third, the sequences associated with the words were cleaned, and ambiguities resolved. After these tasks were completed, the words were ranked by the combined score and included in the peptide database based on empirically determined thresholds. Each unique sequence per abstract identified by text mining was assigned one database entry. Multiple occurrences of the same sequence in different forms, such as 'RGD' and 'Arg-Gly-Asp', were considered a single entry.

Text mining was performed on a Fedora Core 5 Linux virtual machine running on an HP DL320 server with two 3 GHz Xeon processors, allocated 512 MB of RAM. The data resided on a file server connected via Gigabit Ethernet. Text mining of the entire MEDLINE (baseline distribution and updates) took 44 hours, with an additional 16 hours for pre-processing: downloading, uncompressing/compressing and parsing MEDLINE distribution files. The resulting database was 35 MB. Incremental weekly processing of MEDLINE updates took on average under 1 hour.

Step 1. Classification of abstracts

MEDLINE entries that were either duplicates, or did not have abstracts, or were older than 1950 were removed. The older abstracts, which were published prior to the development of Edman degradation 52, did not contain peptide sequences. Several pattern categories of interest were created, such as those related to peptides, phage display, proteases, and others. For each abstract, the total number of matches to patterns in each category was computed, for example, for the 'peptide' category this included the number of matches to 'peptid' or 'hormone', and if at least one of these patterns was present, additionally included the number of matches to less specific patterns such as 'sequenc' or 'motif'. The title, abstract, medical subject heading terms and the chemical list were all scored. Some of the abstracts, especially those published before mid-1990s, often include peptide sequences which are related to protein digestion and sequencing. These sequences usually represent parts of longer proteins, rather than individual peptides, and were thus scored differently. Any matches to this 'digestion' category of patterns were counted. The abstract score was computed as the sum of the number of matches to categories 'peptide' and 'phage', minus the number of matches to the 'digestion' category. Additional terms were added to the abstract score for matches to more than one pattern category in the same abstract, for example the number of matches to patterns from the 'phage' category multiplied by the number of matches to the 'peptide' category. Phosphorylated peptides, such as those selected using the oriented phosphopeptide library technique 53, were not scored any differently from other peptides, that is, neither included not excluded specifically. There is a useful resource, Scansite, dedicated specifically to the phosphorylated peptides 36, which can be used for this application. Texts with a large number or fraction of words in all caps tend to produce many false positives, thus the abstract score was decreased for such abstracts. The abstract score was then transformed for convenience to the (0,1) interval using the function: y = x/(1+x). An abstract score below 0 was assigned to 0. An abstract related to peptide sequences tended to have a score close to 1, and an unrelated one to 0.

Step 2. Classification of words

Each abstract was split into words on whitespace. Each word was matched against a series of peptide sequence pattern categories, in order of decreasing specificities of patterns, until the first successful match. The pattern categories were: full names of amino acids (longest, most specific, such as 'valine' or 'valyl'), 3 letter symbols (such as 'Val') and 1 letter symbols (such as 'V', least specific). Because the recommendations of IUPAC-IUB for reporting peptide sequences 14 were not followed in a large number of abstracts, we had to use a complex classification method and added methods to clean sequences and resolve the ambiguities. Any word that matched a pattern of peptide sequence of at least two amino acids was assigned a score. The score was an empirically calculated measure used to distinguish peptide sequences from other terms, such as nucleic acid sequences, gene symbols, acronyms and all caps English words, which they sometimes closely resemble or are even identical to, when taken out of context.

The above score was defined by several factors. The length/amino acid symbol factor was based on the length of the sequence in amino acids (higher score for longer sequence patterns, which were more specific) and on the type of amino acid symbols used (higher score for the more specific full names than for 1 letter symbols). The degenerate amino acid factor was based on the fraction and the total number of degenerate amino acids (lower score for degenerate amino acids such as 'X' or 'Xaa', which may represent, for example, the starting randomized phage display library rather than the selected peptide). Other factors reflected similarity to either of the following categories: Roman numerals, nucleic acid sequences, gene names and gene symbols, English words, scientific terms or abbreviations, or a combination of the above. The list of abbreviations was derived from the comprehensive ADAM database 54. The list of gene names and symbols was derived from Entrez Gene 55, UniProt 1 and Human Gene Nomenclature (HGNC) 56 databases. An additional factor represented similarity of a given word to protein sequences relative to English words. It was computed for all words that matched a pattern of sequences in 1 letter amino acid symbols. The word was broken up into overlapping k-mers. For example, for k = 3, word 'EYHHYNK' was broken up into 'EYH', 'YHH', 'HHY', 'HYN', 'YNK'. The proportions of all possible k-mers were precomputed in the databases of known protein sequences (from UniProt) and non-sequences (here, English words from MEDLINE abstracts not related to peptides), designated P_pand P_n, respectively. We used the databases of protein sequences and non-sequences of 8 × 10⁷k-mers each, with k = 3, replacing counts of 0 with 1 to avoid division by 0. The protein/English word similarity factor was defined as the product over all overlapping k-mers within the word of (P_p/P_n). For a word with all k-mers equally frequent among sequences and non-sequences, the factor was 1, while for a word such as 'EYHHYNK' in which on average the k-mers were more frequent in protein sequences than in English words, the factor was greater than 1.

The word score was transformed to the (0,1) interval, similarly as in the abstract score. The word score thus depended only on the properties of the word itself, rather than on the context (the properties of the abstract). The combined word/abstract score was then computed for each word, and reflected the abstract score, the word score, and the maximum word score for all words in the abstract, included because sequences tend to occur together in abstracts. The combined word/abstract score s_cwas computed according to the formula

s_c= s_a(w₁s_w+ w₂s_m), for s_w> 0,

s_c= 0, for s_w= 0,

where s_ais the abstract score, s_wis the word score of the current word, s_mis the maximum word score for all words in the abstract, and w₁, w₂are the weights (w₁> w₂). The combined score varied in the (0,1) interval. Words that matched peptide sequence patterns in abstracts related to peptides tended to have a score close to 1, and close to 0 otherwise.

Step 3. Clean-up

Words that matched peptide sequence patterns were cleaned in a series of steps and converted to 1 letter amino acid symbols, as follows. The terminal marks and modifications, such as 'H(2)N-' or '-CO-Ph', were removed. Numbers representing amino acid positions were removed. Other modifications, such as phosphate in 'pY' were removed. Motifs such as '(L/I)' or 'L/I' were resolved. Amino acids that do not have a 1 letter IUPAC symbol were replaced with X. As a result, a large variety of different sequence formats were resolved, including 'N-acetyl-l-aspartyl-l-glutamyl-l-valyl-l-aspartyl-7-amino-4-methylcoumarin' to 'DEVD', 'Gly1-Val2-Thr3-Ser4' to 'GVTS', '(Arg-Glu(EDANS)-Ser-Gln)' to 'RESQ', 'TRDI-pY-ETD-pY-pY-RK' to 'TRDIYETDYYRK', and others.

To estimate precision of text mining, 50 sequences with the combined score above the threshold for inclusion in PepBank were selected at random from the text mining output. Each of these positive predictions was manually verified, whether or not the word contained a peptide sequence (40 out of 50 were found correctly, precision = 0.8), and whether or not the word contained a peptide sequence AND the sequence was parsed 100% correctly (35 out of 50 correct, precision = 0.7). If the identified sequence was a partial protein sequence, rather than a peptide or a phage display sequence, it was considered an error: such sequences are typically entered in protein databases and do not need to be mined from text (most of the errors in precision were of this type). One or more incorrect amino acid was also considered an error.

For estimating recall, we created a separate test set of 50 sequences by searching in PubMed for recent review articles using as a query "peptide OR peptides" alone or in combination with "sequence OR sequences", and followed the PubMed abstract links for the references cited in the reviews. Peptide sequences were manually extracted from the abstracts without any automated pattern matching. The text mining output with the combined score above the threshold for inclusion in PepBank was matched against these positive real cases. Again, for each case we manually verified whether or not the algorithm found the word, which contained this peptide sequence (12 out of 50 correct, recall = 0.24), and whether or not the algorithm found the word AND the sequence was parsed 100% correctly (10 out of 50 correct, recall = 0.2). Most of the errors in recall were due to blanks (often typos) inside peptide sequences or due to unrecognized amino acid modifications.

The pioneering method to identify DNA and protein sequences in text, based on Markov models was described by Wren and co-workers 57. Our text mining method, while similar in spirit, has different goals and thus uses a different sequence identification strategy. One of our main goals was to rapidly identify peptides with potential therapeutic and diagnostic utility (including those derived from phage display peptides), rather than identifying peptide epitopes and providing an aid to their manual curation. We also use extensive context information from the abstract, and collect peptide motifs in addition to sequences. We clean the sequences and provide access to the data for biologists through a simple web-based interface for text and sequence similarity searches. We do not place a minimum length restriction on sequences, such as 6 amino acids, because many therapeutic peptides are relatively short, for example the well-known RGD motif and many others found in phage display. Due to the substantial differences in goals and methods between our approach and that of others, it may be interesting to develop in the future a hybrid method combining the strengths of both approaches.

Other sources

All peptide sequences with length 20 or below were extracted from ASPD 11 and UniProt 1, and fields that mapped to PepBank were parsed and stored (for example, interactor fields from ASPD, peptide fields from UniProt). The links from PepBank to the source databases were provided for all entries. Many of the peptides were stored in UniProt as part of the longer precursor proteins, producing peptides on cleavage. These peptide sequences were extracted using the UniProt feature table by selecting those with feature key "peptide" or "chain" and feature length under 20. Additional entries were manually curated, capturing the available interaction data, from the full text articles on phage display in PDF format. The articles were chosen to represent a small but diverse selection of reports within this field.

Utility and discussion

User interface

The web-based user interface to PepBank offers text search (both Quick and Advanced), as well as sequence similarity search (BLAST and Smith-Waterman algorithms). The Quick Search function offers a simple, Google-like search for biologists looking for peptide data in all fields. Advanced Search options include querying data by individual fields. Exact search, wildcard (*) and any single character (_) are supported in text search, which enables, for example, searching for a sequence pattern as a query. The results of the text search are displayed as a table sortable in the browser, with hyperlinks to the original sources (MEDLINE/PubMed, ASPD, UniProt) and to more detailed information.

Text search example: VEGFR related peptides

To illustrate the utility of PepBank, we use the example of identifying peptides with affinity to VEGFR1, an important therapeutic target 58. The user can search for VEGFR using either Quick or Advanced Search, obtain a set of peptide sequences related to this target, and view details for the selected sequences. In the example shown in Figure 2, sequence 'WHSDMEWWYLLG' is identified 59. Prompted by these results, the user of PepBank may be interested in testing this peptide sequence in novel forms (for example, dendrimers, or conjugated to nanoparticles), or for novel biomedical applications (imaging different tumor types, atherosclerosis, or arthritis). There is currently no database where the user can easily obtain such information as it relates to molecular targets and peptide sequences. One can also query directly for a biological process (such as apoptosis or angiogenesis) or for the target cell line or tissue (such as BICR-H1 or U937).

Figure 2

Web-based user interface of PepBank

Web-based user interface of PepBank. Illustration of a typical user workflow. The user enters the query with Quick or Advanced Search. The results are returned in a table sortable in the browser. The user selects the entry or entries of interest. The sequence in the example shown was obtained by text mining and was then manually curated. The score, between 0 and 1, reflects the degree of confidence in the interaction (higher score for more confidence). Manually curated entries receive higher score than entries from automated text mining.

To determine whether the database would yield target leads against known drug targets, we randomly chose a set of 20 defined drug targets from the 547 approved drug target data set in DrugBank 60. The randomly chosen drug targets were not skewed towards peptide receptors and included: squalene epoxidase, RAF proto-oncogene serine/threonine-protein kinase, muscarinic acetylcholine receptor M4, opioid mu receptor (OP3), adenosine A1 receptor, GABA transaminase, amidophosphoribosyltransferase precursor, tryptophan 5-hydroxylase 1, apoptosis regulator Bcl-2, matrix protein M2, vascular endothelial growth factor receptor 2 precursor, amiloride-sensitive sodium channel gamma-subunit, ribonucleotide reductase, cAMP phosphodiesterase, coagulation factor VIII, high affinity immunoglobulin epsilon receptor alpha-subunit precursor, retinol-binding protein I, glycine alpha 2 receptor, cytochrome P450 51, GABA-A receptor subunit (C. elegans). Relevant peptides were defined as those interacting with the target or its ortholog, or modulating the function of the target, for example by acting as a competitor. Relevant peptides in our database were identified in approximately 25% of the above drug targets.

Sequence similarity search examples

As an illustrative example, we performed an all-against-all BLAST search of PepBank sequences. One of the surprises was the discovery of an exact match to sequence 'GETRAPL' from phage display selection for peptides that bind to secreted protein acidic and rich in cysteine (SPARC) 61. The sequence had a BLAST hit with an E-value of 0.06 to an isolate from phage display selection of peptides that bind human saphenous vein smooth muscle cells 62. Following the BLAST results, we then found that in addition to these 2 selections, the exact same sequence was isolated independently multiple times by different groups in selections with unrelated targets. GETRAPL was found in phage display selections of peptides that bind human immunodeficiency virus type 1 (HIV-1) accessory viral protein (Vpr) 63, chromatin high mobility group protein 1, box A (HMGB1) from rat 64, mouse skeletal muscle tissue in vivo 65, and mouse brain cells in vivo 66.

We suggest that one of the utilities for PepBank is to search the peptide sequences of interest to the user with BLAST or Smith-Waterman algorithms to find any important similarities to the known peptides collected in our database. In this example, the search can be used to remove a relatively nonspecific binder GETRAPL. Note that searching PepBank with these tools is a unique resource: an exact match may be easy to find, but using a partial match such as GETRA as a query finds GETRAPL only in PepBank, but not in PubMed 2 or on Google. Searching with BLAST 67 or with Smith-Waterman/SSEARCH methods 47 using GETRAPL as a query against nr database 2 gives no peptide hits cited above. A large interactions database IntAct 6 gives no hits for GETRAPL query at all.

Another surprise discovery in the all-against-all BLAST search of PepBank sequences was the multiple occurrence of the sequence SVSVGMKPSPRP. The sequence had several exact matches over its entire length of 12 amino acids, with an E-value of 1 × 10^-6. It was isolated in phage display selection for peptides that bind to DNA 68. In this selection SVSVGMKPSPRP was the only sequence studied due to its dominance (9 out of 10) in the selected pool. The exact same sequence was isolated in phage display selection for peptides binding to human monoclonal IgM 69, and to the mirror image of Alzheimer's disease amyloid peptide Abeta(1–42) 70. The sources for these sequences were MEDLINE abstract text mining, ASPD database, and manually curated full text articles, respectively. In addition, SVSVGMKPSPRP occurs in several patents 7172. Several groups note multiple isolation of this remarkable sequence in their own and other, unrelated, experiments 7374. The sequence has also been identified in a recent excellent review 24 which covers the important topic of target-unrelated sequences in phage display. Interestingly, all of the studies with both GETRAPL and SVSVGMKPSPRP were done with the phage display libraries from the same manufacturer, thus suggesting a library- or methodology-specific phenomenon. Both sequences illustrate one of the suggested utilities for PepBank, namely that one can search it with a sequence query using BLAST or Smith-Waterman algorithms to find any important similarities to the known peptides.

Conclusion

A new text mining tool was developed and used to identify peptide sequences in MEDLINE abstracts. These data were combined with two of the public sources of peptide sequence data, ASPD and UniProt, as well as with manually curated peptide data. The database application was developed to query the data using text and sequence similarity search through a web-based user interface. The utility of PepBank was demonstrated using different examples of peptide sequences. The results show that the database has valuable biological and medical applications. In the future, we plan to add other public sources of peptide data, such as the peptide subset of the Molecular Interaction database (MINT) 5, and other sources for text mining, such as full-text journal articles. Also, in the future we will apply machine learning techniques to improve the accuracy of text mining to extract sequences. In the next release, we plan to add the functionalities to download the data in a standard format, such as PSI MI, and to search the database for peptide motifs.

Availability and requirements

The database is freely available on http://pepbank.mgh.harvard.edu/, and the text mining source code (Peptide::Pubmed) is freely available above as well as on CPAN http://www.cpan.org/.

Authors' contributions

TS designed and developed the text mining algorithm, curated the database contents, co-designed the database and the interface and wrote the manuscript, DG designed and developed the database, the web application and the interface, MK co-curated the database contents, MP designed the architecture of the entire web site and designed the database and the interface, RW provided the conceptual design and the overall guidance of the entire project and co-wrote the manuscript. All authors read and approved the final manuscript.

Acknowledgements

We thank Timo Duchrow, Vladimir Kubatin, Lee Josephson, Ching Tung, Elena Aikawa, Kim Kelly and Rajesh Anbazhagan for helpful discussions and feedback on the database, Jason Brown and Brett Dikeman for system administration work and Melissa Carlson for editorial assistance. We are grateful to the authors and curators of the resources we used: ADAM (in particular, Neil Smalheiser), MEDLINE/NLM, UniProt and ASPD, and to anonymous reviewers for their comments. This work was supported in part by NIH grants PO1-AI54904 (RW), P50-CA86355 (RW), U54-CA126515 (RW).

The Universal Protein Resource (UniProt): an expanding universe of protein information Wu CH Apweiler R Bairoch A Natale DA Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M Martin MJ Mazumder R O'Donovan C Redaschi N Suzek B Nucleic Acids Res 2006 34 Database issue D187 91 1347523 16381842 10.1093/nar/gkj161 Database resources of the National Center for Biotechnology Information Wheeler DL Barrett T Benson DA Bryant SH Canese K Chetvernin V Church DM DiCuccio M Edgar R Federhen S Geer LY Kapustin Y Khovayko O Landsman D Lipman DJ Madden TL Maglott DR Ostell J Miller V Pruitt KD Schuler GD Sequeira E Sherry ST Sirotkin K Souvorov A Starchenko G Tatusov RL Tatusova TA Wagner L Yaschenko E Nucleic Acids Res 2007 35 Database issue D5 12 1781113 17170002 10.1093/nar/gkl1031 The International Protein Index: an integrated database for proteomics experiments Kersey PJ Duarte J Williams A Karavidopoulou Y Birney E Apweiler R Proteomics 2004 4 7 1985 1988 10.1002/pmic.200300721 15221759 The Biomolecular Interaction Network Database and related tools 2005 update Alfarano C Andrade CE Anthony K Bahroos N Bajec M Bantoft K Betel D Bobechko B Boutilier K Burgess E Buzadzija K Cavero R D'Abreo C Donaldson I Dorairajoo D Dumontier MJ Dumontier MR Earles V Farrall R Feldman H Garderman E Gong Y Gonzaga R Grytsan V Gryz E Gu V Haldorsen E Halupa A Haw R Hrvojic A Hurrell L Isserlin R Jack F Juma F Khan A Kon T Konopinsky S Le V Lee E Ling S Magidin M Moniakis J Montojo J Moore S Muskat B Ng I Paraiso JP Parker B Pintilie G Pirone R Salama JJ Sgro S Shan T Shu Y Siew J Skinner D Snyder K Stasiuk R Strumpf D Tuekam B Tao S Wang Z White M Willis R Wolting C Wong S Wrong A Xin C Yao R Yates B Zhang S Zheng K Pawson T Ouellette BF Hogue CW Nucleic Acids Res 2005 33 Database issue D418 24 540005 15608229 10.1093/nar/gki051 MINT: the Molecular INTeraction database Chatr-aryamontri A Ceol A Palazzi LM Nardelli G Schneider MV Castagnoli L Cesareni G Nucleic Acids Res 2007 35 Database issue D572 4 1751541 17135203 10.1093/nar/gkl950 IntAct--open source resource for molecular interaction data Kerrien S Alam-Faruque Y Aranda B Bancarz I Bridge A Derow C Dimmer E Feuermann M Friedrichsen A Huntley R Kohler C Khadake J Leroy C Liban A Lieftink C Montecchi-Palazzi L Orchard S Risse J Robbe K Roechert B Thorneycroft D Zhang Y Apweiler R Hermjakob H Nucleic Acids Res 2007 35 Database issue D561 5 1751531 17145710 10.1093/nar/gkl958 MIPS: analysis and annotation of proteins from whole genomes in 2005 Mewes HW Frishman D Mayer KF Munsterkotter M Noubibou O Pagel P Rattei T Oesterheld M Ruepp A Stumpflen V Nucleic Acids Res 2006 34 Database issue D169 72 1347510 16381839 10.1093/nar/gkj148 Human protein reference database--2006 update Mishra GR Suresh M Kumaran K Kannabiran N Suresh S Bala P Shivakumar K Anuradha N Reddy R Raghavan TM Menon S Hanumanthu G Gupta M Upendran S Gupta S Mahesh M Jacob B Mathew P Chatterjee P Arun KS Sharma S Chandrika KN Deshpande N Palvankar K Raghavnath R Krishnakanth R Karathia H Rekha B Nayak R Vishnupriya G Kumar HG Nagini M Kumar GS Jose R Deepthi P Mohan SS Gandhi TK Harsha HC Deshpande KS Sarker M Prasad TS Pandey A Nucleic Acids Res 2006 34 Database issue D411 4 1347503 16381900 10.1093/nar/gkj141 DIP, the Database of Interacting Proteins: a research tool for studying cellular networks of protein interactions Xenarios I Salwinski L Duan XJ Higney P Kim SM Eisenberg D Nucleic Acids Res 2002 30 1 303 305 99070 11752321 10.1093/nar/30.1.303 RELIC--a bioinformatics server for combinatorial peptide analysis and identification of protein-ligand interaction sites Mandava S Makowski L Devarapalli S Uzubell J Rodi DJ Proteomics 2004 4 5 1439 1460 10.1002/pmic.200300680 15188413 ASPD (Artificially Selected Proteins/Peptides Database): a database of proteins and peptides evolved in vitro Valuev VP Afonnikov DA Ponomarenko MP Milanesi L Kolchanov NA Nucleic Acids Res 2002 30 1 200 202 99101 11752292 10.1093/nar/30.1.200 Synthetic antibiotic peptides database Wade D Englund J Protein Pept Lett 2002 9 1 53 57 10.2174/0929866023408986 12141924 APD: the Antimicrobial Peptide Database Wang Z Wang G Nucleic Acids Res 2004 32 Database issue D590 2 308759 14681488 10.1093/nar/gkh025 IUPAC-IUB Joint Commission on Biochemical Nomenclature (JCBN). Nomenclature and symbolism for amino acids and peptides. Recommendations 1983 Biochem J 1984 219 2 345 373 1153490 6743224 Amino Acids and Peptides Barrett JC Elmore DT Cambridge , Cambridge University Press 1998 Dosimetry in Peptide radionuclide receptor therapy: a review Cremonesi M Ferrari M Bodei L Tosi G Paganelli G J Nucl Med 2006 47 9 1467 1475 16954555 Candidates for peptide receptor radiotherapy today and in the future Reubi JC Macke HR Krenning EP J Nucl Med 2005 46 Suppl 1 67S 75S 15653654 High- versus low-dose oxytocin for augmentation or induction of labor Patka JH Lodolce AE Johnston AK Ann Pharmacother 2005 39 1 95 101 15572602 Protamine as an efficient membrane-translocating peptide Reynolds F Weissleder R Josephson L Bioconjug Chem 2005 16 5 1240 1245 10.1021/bc0501451 16173804 Detection of vascular adhesion molecule-1 expression using a novel multimodal nanoparticle Kelly KA Allport JR Tsourkas A Shinde-Patil VR Josephson L Weissleder R Circ Res 2005 96 3 327 336 10.1161/01.RES.0000155722.17881.dd 15653572 A novel method for imaging apoptosis using a caspase-1 near-infrared fluorescent probe Messerli SM Prabhakar S Tang Y Shah K Cortes ML Murthy V Weissleder R Breakefield XO Tung CH Neoplasia 2004 6 2 95 105 1502090 15140398 10.1593/neo.03214 Peptidomics: the comprehensive analysis of peptides in complex biological mixtures Schulz-Knappe P Zucht HD Heine G Jurgens M Hess R Schrader M Comb Chem High Throughput Screen 2001 4 2 207 217 11281836 In vivo interrogation of the molecular display of atherosclerotic lesion surfaces Liu C Bhattacharjee G Boisvert W Dilley R Edgington T Am J Pathol 2003 163 5 1859 1871 1892421 14578186 The nature of target-unrelated peptides recovered in the screening of phage-displayed random peptide libraries with antibodies Menendez A Scott JK Anal Biochem 2005 336 2 145 157 1464800 15620878 10.1016/j.ab.2004.09.048 Basic local alignment search tool Altschul SF Gish W Miller W Myers EW Lipman DJ J Mol Biol 1990 215 3 403 410 2231712 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Nucleic Acids Res 1997 25 17 3389 3402 146917 9254694 10.1093/nar/25.17.3389 Identification of common molecular subsequences Smith TF Waterman MS J Mol Biol 1981 147 1 195 197 10.1016/0022-2836(81)90087-5 7265238 Searching protein sequence libraries: comparison of the sensitivity and selectivity of the Smith-Waterman and FASTA algorithms Pearson WR Genomics 1991 11 3 635 650 10.1016/0888-7543(91)90071-L 1774068 Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface Smith GP Science 1985 228 4705 1315 1317 10.1126/science.4001944 4001944 Phage Display Smith GP Petrenko VA Chem Rev 1997 97 2 391 410 10.1021/cr960065d 11848876 Small molecules, big players: the National Cancer Institute's Initiative for Chemical Genetics Tolliday N Clemons PA Ferraiolo P Koehler AN Lewis TA Li X Schreiber SL Gerhard DS Eliasof S Cancer Res 2006 66 18 8935 8942 10.1158/0008-5472.CAN-06-2552 16982730 The role of structure in antibody cross-reactivity between peptides and folded proteins Craig L Sanschagrin PC Rozek A Lackie S Kuhn LA Scott JK J Mol Biol 1998 281 1 183 201 10.1006/jmbi.1998.1907 9680484 Designing scaffolds of peptides for phage display libraries Uchiyama F Tanaka Y Minari Y Tokui N J Biosci Bioeng 2005 99 5 448 456 10.1263/jbb.99.448 16233816 PenBase, the shrimp antimicrobial peptide penaeidin database: sequence-based classification and recommended nomenclature Gueguen Y Garnier J Robert L Lefranc MP Mougenot I de Lorgeril J Janech M Gross PS Warr GW Cuthbertson B Barracco MA Bulet P Aumelas A Yang Y Bo D Xiang J Tassanakajon A Piquemal D Bachere E Dev Comp Immunol 2006 30 3 283 288 10.1016/j.dci.2005.04.003 15963564 Defensins knowledgebase: a manually curated database and information source focused on the defensins family of antimicrobial peptides Seebah S Suresh A Zhuo S Choong YH Chua H Chuon D Beuerman R Verma C Nucleic Acids Res 2007 35 Database issue D265 8 1669742 17090586 10.1093/nar/gkl866 Scansite 2.0: Proteome-wide prediction of cell signaling interactions using short sequence motifs Obenauer JC Cantley LC Yaffe MB Nucleic Acids Res 2003 31 13 3635 3641 168990 12824383 10.1093/nar/gkg584 SYFPEITHI: database for MHC ligands and peptide motifs Rammensee H Bachmann J Emmerich NP Bachor OA Stevanovic S Immunogenetics 1999 50 3-4 213 219 10.1007/s002510050595 10602881 EPIMHC: a curated database of MHC-binding peptides for customized computational vaccinology Reche PA Zhang H Glutting JP Reinherz EL Bioinformatics 2005 21 9 2140 2141 10.1093/bioinformatics/bti269 15657103 MHCBN: a comprehensive database of MHC binding and non-binding peptides Bhasin M Singh H Raghava GP Bioinformatics 2003 19 5 665 666 10.1093/bioinformatics/btg055 12651731 JenPep: a database of quantitative functional peptide data for immunology Blythe MJ Doytchinova IA Flower DR Bioinformatics 2002 18 3 434 439 10.1093/bioinformatics/18.3.434 11934742 MPID: MHC-Peptide Interaction Database for sequence-structure-function information on peptides binding to MHC molecules Govindarajan KR Kangueane P Tan TW Ranganathan S Bioinformatics 2003 19 2 309 310 10.1093/bioinformatics/19.2.309 12538264 Population of the HLA ligand database Sathiamurthy M Hickman HD Cavett JW Zahoor A Prilliman K Metcalf S Fernandez Vina M Hildebrand WH Tissue Antigens 2003 61 1 12 19 10.1034/j.1399-0039.2003.610102.x 12622773 The HUPO PSI's molecular interaction format--a community standard for the representation of protein interaction data Hermjakob H Montecchi-Palazzi L Bader G Wojcik J Salwinski L Ceol A Moore S Orchard S Sarkans U von Mering C Roechert B Poux S Jung E Mersch H Kersey P Lappe M Li Y Zeng R Rana D Nikolski M Husi H Brun C Shanker K Grant SG Sander C Bork P Zhu W Pandey A Brazma A Jacq B Vidal M Sherman D Legrain P Cesareni G Xenarios I Eisenberg D Steipe B Hogue C Apweiler R Nat Biotechnol 2004 22 2 177 183 10.1038/nbt926 14755292 Ruby on Rails http://www.rubyonrails.org MySQL http://www.mysql.com The National Center for Biotechnology Information (NCBI) ftp site ftp://ftp.ncbi.nih.gov/ University of Virginia FASTA server http://fasta.bioch.virginia.edu/fasta_www2/fasta_list2.shtml The National Library of Medicine (NLM) ftp site ftp://ftp.nlm.nih.gov/ CGMIM: automated text-mining of Online Mendelian Inheritance in Man (OMIM) to identify genetically-associated cancers and candidate genes Bajdik CD Kuo B Rusaw S Jones S Brooks-Wilson A BMC Bioinformatics 2005 6 78 1274267 15796777 10.1186/1471-2105-6-78 Text mining neuroscience journal articles to populate neuroscience databases Crasto CJ Marenco LN Migliore M Mao B Nadkarni PM Miller P Shepherd GM Neuroinformatics 2003 1 3 215 237 10.1385/NI:1:3:215 15046245 PreBIND and Textomy--mining the biomedical literature for protein-protein interactions using a support vector machine Donaldson I Martin J de Bruijn B Wolting C Lay V Tuekam B Zhang S Baskin B Bader GD Michalickova K Pawson T Hogue CW BMC Bioinformatics 2003 4 11 153503 12689350 10.1186/1471-2105-4-11 Method for determination of the amino acid sequence in peptides Edman P Acta Chem Scand 1950 4 283 293 SH2 domains recognize specific phosphopeptide sequences Songyang Z Shoelson SE Chaudhuri M Gish G Pawson T Haser WG King F Roberts T Ratnofsky S Lechleider RJ Cell 1993 72 5 767 778 10.1016/0092-8674(93)90404-E 7680959 ADAM: another database of abbreviations in MEDLINE Zhou W Torvik VI Smalheiser NR Bioinformatics 2006 22 22 2813 2818 10.1093/bioinformatics/btl480 16982707 Entrez Gene: gene-centered information at NCBI Maglott D Ostell J Pruitt KD Tatusova T Nucleic Acids Res 2007 35 Database issue D26 31 1761442 17148475 10.1093/nar/gkl993 The HUGO Gene Nomenclature Database, 2006 updates Eyre TA Ducluzeau F Sneddon TP Povey S Bruford EA Lush MJ Nucleic Acids Res 2006 34 Database issue D319 21 1347509 16381876 10.1093/nar/gkj147 Markov model recognition and classification of DNA/protein sequences within large text databases Wren JD Hildebrand WH Chandrasekaran S Melcher U Bioinformatics 2005 21 21 4046 4053 10.1093/bioinformatics/bti657 16159926 Development of vascular endothelial growth factor receptor (VEGFR) kinase inhibitors as anti-angiogenic agents in cancer therapy Underiner TL Ruggeri B Gingrich DE Curr Med Chem 2004 11 6 731 745 10.2174/0929867043455756 15032727 Suppression of tumor growth and metastasis by a VEGFR-1 antagonizing peptide identified from a phage display library An P Lei H Zhang J Song S He L Jin G Liu X Wu J Meng L Liu M Shou C Int J Cancer 2004 111 2 165 173 10.1002/ijc.20214 15197767 DrugBank: a comprehensive resource for in silico drug discovery and exploration Wishart DS Knox C Guo AC Shrivastava S Hassanali M Stothard P Chang Z Woolsey J Nucleic Acids Res 2006 34 Database issue D668 72 1347430 16381955 10.1093/nar/gkj067 In vivo imaging of molecularly targeted phage Kelly KA Waterman P Weissleder R Neoplasia 2006 8 12 1011 1018 1783712 17217618 10.1593/neo.06610 Development of efficient viral vectors selective for vascular smooth muscle cells Work LM Nicklin SA Brain NJ Dishart KL Von Seggern DJ Hallek M Buning H Baker AH Mol Ther 2004 9 2 198 208 10.1016/j.ymthe.2003.11.006 14759804 Design and assay of inhibitors of HIV-1 Vpr Cell Killing and growth arrest activity using microbial assay systems Sankovich SE Koleski D Baell J Matthews B Azad AA Macreadie IG J Biomol Screen 1998 3 4 299 304 10.1177/108705719800300409 HMGB1 interacts with many apparently unrelated proteins by recognizing short amino acid sequences Dintilhac A Bernues J J Biol Chem 2002 277 9 7021 7028 10.1074/jbc.M108417200 11748221 Methods and compositions for targeting compounds to muscle. United States Patent 6399575 Smith BF Samoilova T 2001 Methods and compositions for targeting compounds to the central nervous system. United States Patent 6399575 Smith BF Samoilova T Baker HJ 2002 The National Center for Biotechnology Information (NCBI) BLAST server http://www.ncbi.nlm.nih.gov/BLAST/ A DNA-binding peptide from a phage display library Wolcke J Weinhold E Nucleosides Nucleotides Nucleic Acids 2001 20 4-7 1239 1241 10.1081/NCN-100002526 11562993 Two human neonatal IgM antibodies encoded by different variable-region genes bind the same linear peptide: evidence for a stereotyped repertoire of epitope recognition Messmer BT Sullivan JJ Chiorazzi N Rodman TC Thaler DS J Immunol 1999 162 4 2184 2192 9973494 Selection of D-amino-acid peptides that bind to Alzheimer's disease amyloid peptide abeta1-42 by mirror image phage display Wiesehan K Buder K Linke RP Patt S Stoldt M Unger E Schmitt B Bucci E Willbold D Chembiochem 2003 4 8 748 753 10.1002/cbic.200300631 12898626 Control of crop pests & animal parasites through direct neuronal uptake. United States Patent 20030181376 Atkinson HJ McPherson MJ Winter MD 2003 Identification of peptides that facilitate uptake and cytoplasmic and/or nuclear transport of proteins, DNA and viruses. United States Patent 20030219826 Robbins PD Mi Z Frizzell R Glorioso JC Gambotto A Mai JC 2003 A novel method to identify and characterise peptide mimotopes of heat shock protein 70-associated antigens Arnaiz B Madrigal-Estebas L Todryk S James TC Doherty DG Bond U J Immune Based Ther Vaccines 2006 4 2 1482705 16603084 10.1186/1476-8518-4-2 Selection by phage display of peptides targeting the HIV-1 TAR element Kolb G Boiziau C RNA Biol 2005 2 1 28 33 17132933