The Proteomics Standards Initiative session of the Human Proteome Organization (HUPO) 2005 congress was opened by the current chair, Rolf Apweiler of the European Molecular Biology Laboratory - European Bioinformatics Institute (EMBL-EBI), who summarized the achievements of the HUPO Proteomics Standards Initiative (HUPO-PSI) to date. The Molecular Interaction XML interchange standard is already widely used, and all of the major publicly available databases now make data available in this format [1].
Five of these databases -- BIND, DIP, IntAct, MINT, and MPact (MIPS) -- have formed the International Molecular Exchange consortium to jointly curate and exchange data, with data exchange commencing early in 2006. The mass spectrometry standard (PSI-MS mzData version 1.05) for the collection of raw peaklist data is being implemented by an increasing number of instrumentation and software manufacturers [2]. The Institute for Systems Biology in Seattle has also released a second standard that was developed for use in their analysis pipeline [3]. The two groups have agreed to work together to ensure that first the analysis standard (analysisXML) and later the two peaklist standards will be merged into a single entity to facilitate user adoption.
Henning Hermjakob of EMBL-EBI detailed the current status of the PSI-MI standard version 2.5, which has been frozen in preparation for release in September 2005. The standard has already been successfully implemented by one database, and others have committed to release within a few weeks of publication. It is planned that PSI-MI2.5 will be the supported standard from January 2006 and will be supported by both the current suite of tools and new applications such as PSIQUIC, a common query Web service API for all participating databases.
Hermjakob went on to describe work undertaken by Luisa Montecchi-Palazzi and John Garavelli of EMBL-EBI in developing an ontology of terms to describe protein modifications, both the natural and artifactual modifications observed by mass spectrometrists. This work will combine the modifications collected by three databases -- RESID, UniMod, and Delta Mass -- and the controlled vocabulary already developed for PSI-MI by Luisa Montecchi-Palazzi into a single resource for all proteomics studies.
The mass spectrometry group is now co-chaired by Randall Julian of Eli Lilly and Patrick Pedrioli of ISMB, Zurich. However, since neither was present at this pre-congress session, Chris Taylor of EMBL-EBI briefly updated the delegates with a list of manufacturers who have either already incorporated mzData v.1.05 into their procedures or are planning to do so on their next release. Merger with the ISB standard will have to be carefully planned, since a stable platform is vital to ensure vendor compliance.
One immediate benefit of the mzData standard is that the data generated by the HUPO tissue initiatives is now available in the PRIDE mzData-compatible database from where it can be accessed and downloaded for subsequent reanalysis. An initial meeting of the major current, and planned, proteomics mass spectrometry repositories was scheduled for immediately after the main congress to discuss data exchange. Taylor went on to describe the continuing efforts to describe a General Proteomics Standard. This group is now working with the FuGE (Functional Genomics Experiment) efforts to define a general standard in which to encode data, which will enable a systems biology approach to data analysis. Immediate future plans center on the development of standards for the description, storage, and exchange of gel electrophoresis data that are to developed over the next few months.
Rudi Aebersold of ISB, Swiss Federal Institute of Technology, Zurich, the chair-elect of the Bioinformatics committee, ended the session by praising the amazing progress made by the different groups and their already considerable contributions to the HUPO tissue initiatives. All the different strands are actively pushing these standards forward and ensuring their adoption by the scientific community and the associated instrumentation manufacturers.
All such efforts require support from the user community, and the HUPO-PSI is actively seeking input and advice from all quarters. Anyone wishing to become involved is invited to visit http://psidev.sf.net/ to participate in the discussion groups listed and to contribute to the further development of community standards for proteomics data.
Proteomics101: Data Profusion
The field of proteomics is increasingly broad and encompasses an expanding number of complex experimental techniques. Studies generate huge volumes of data that can include gel images, mass spectrometry spectra, and protein identifications, all of which poses enormous problems of analysis, interpretation, and storage for the original researcher.
The ability of researchers to vary the type of instrumentation and analytical methodology used from experiment to experiment only adds to the problem. It is not unusual for a proteomics group to possess more than one type of mass spectrometer and vary their choice of machine with the type of experiment. This can create problems for data comparison even within a single laboratory, since each instrument manufacturer typically provides its own proprietary program to analyze the output.
Publication of this data presents a further challenge. Currently, a typical paper describing a proteomics data set will consist of an experimental description, a summary data table, and an analysis and discussion of the data in that table. Additional supporting material is often supplied as a supplementary table, where it is not available in a format appropriate for reanalysis by a third party. Descriptions of the sample acquisition, preparation, and storage vary enormously, although these are crucial issues that have a direct impact on the final data set. Even an issue as fundamental as protein identity can often be confusing, with author-derived nomenclature failing to coincide with that used by the public-domain databases and accession numbers not being quoted.
The need to develop common data representation standards in the field of proteomics is apparent, with a requirement for a system that is both stable enough to provide a reliable platform for users, instrumentation manufacturers, and software developers to work from and flexible enough to keep pace with new developments in the field of both mass spectrometry and proteomics in general. To address this problem, HUPO [4] set its internal Bioinformatics committee the task of producing standards, initially to enable the transfer of information between laboratories participating in the HUPO tissue initiatives.
References
1 Hermjakob, H. et al. Nat Biotechnol 22, 177-83; 2004.
2 Orchard, S. et al. Proteomics 5, 3552-5; 2005.
3 Pedrioli, P.G. et al. Nat Biotechnol 22, 1459-66; 2004.
4 Kaiser, J. Science 296, 827; 2002.
Sandra Orchard, Henning Hermjakob, and Rolf Apweiler work for the EMBL-EBI. E-mail: orchard@ebi.ac.uk.