Meeting News: Human Protein Atlas unveiled | Proteomics

J. Proteome Res. , 2005, 4 (6), pp 1902–1902. DOI: 10.1021/pr050531s. Publication Date (Web): December 12, 2005. Note: In lieu of an abstract, this ...
0 downloads 0 Views 55KB Size
M E E T I N G

Human Protein Atlas unveiled Hailed as the first deliverable from HUPO, version 1.0 of the Human Protein Atlas was officially launched by the Human Antibody Initiative at HUPO’s annual meeting. Mathias Uhlén and colleagues at the Royal Institute of Technology and Uppsala University (both in Sweden) developed the atlas, which currently contains immunolocalization data on ~700 proteins. Uhlén describes the atlas as “a biomarker discovery program on a very grandiose scale, because we can look at one protein at a time and see if it goes up or down in diseases.” Researchers who visit the freely available resource (www.proteinatlas. org) can browse by chromosome number or keyword. For each protein analyzed, an annotation summary page tabulates the level of its expression by tissue. Clicking on the tissue name brings users to a virtual microscope view of the tissue stained with an antibody to the protein. The Human Protein Atlas currently contains >400,000 images, all of which have been examined by pathologists. Over the past two years, Uhlén has hired 100 researchers with diverse expertise to work on the project. Bioinformaticians determine the best part of a protein to express for antibody production, and molecular biologists make constructs that encode for these protein epitope signature tags (PrESTs). Re-

searchers in the “protein factory” use these constructs to produce large quantities of each PrEST, which are injected into animals. Polyclonal antibodies are harvested and affinity-purified with the original PrEST; this yields monospecific antibodies. To verify an antibody’s specificity, researchers in the quality assurance group test each one on an array of thousands of human proteins. Func-

MATHIAS UHLÉN, ROYAL INSTITUTE OF TECHNOLOGY, SWEDEN

Katie Cottingham reports from the HUPO Fourth Annual World Congress—Munich

Where’s the protein? An antibody to insulin stains cells in the pancreas.

tional biologists screen both antibodies produced by the group and commercially available antibodies on tissue microarrays that contain normal and cancer tissue samples. Finally, pathologists annotate the stained slides. Handling the images is one of the

Proteomics repositories will exchange data On the day after the HUPO 4th Annual World Congress officially closed, a handful of bioinformaticians, journal editors, and scientists remained in Munich for one more meeting to discuss the future of proteomics repositories. Attendees, including representatives of at least five established databases, agreed to work together and develop a mechanism for exchanging their data. According to Rolf Apweiler at the European Bioinformatics Institute, the collaboration (tentatively called ProteomEx) will initially develop an index of data sets to be included in each repository. The index, which participants plan to implement by the end of 2006, will also include journal references and PubMed identifiers for the data sets. Apweiler says that the idea for the index came from using the setup of the IMEx consortium, a group of protein inter-

1902

N E W S

Journal of Proteome Research • Vol. 4, No. 6, 2005

biggest challenges, according to Uhlén. “We are currently producing ~200 gigabytes of data every day,” he says. The researchers, therefore, transfer the images daily to a supercomputer center for storage. Another challenge is managing the workflow. “We tried to use laboratory information management systems [LIMS] that are commercially available, but we went into a wall; they basically just broke down,” says Uhlén. To remedy this situation, the researchers developed their own LIMS. Uhlén plans to make half of the aliquots of his antibodies available to researchers by next spring. The remaining aliquots will be used in collaborations with other HUPO initiatives, such as the Human Plasma Proteome Project and the Human Liver Proteome Project. In addition to producing antibodies, Uhlén says he will also welcome antibodies from the community. “We are planning to [accept antibodies from] external groups. If the [antibodies] work and have good quality, then we will make them available at the Protein Atlas with a reference back to the person who has submitted these antibodies,” he explains. According to Uhlén, the project will keep him and his co-workers busy for a while. He predicts that it will take 10 years to produce antibodies for proteins expressed by every human gene. The next two updates of the Human Protein Atlas will be announced in 2006 at the U.S. HUPO meeting in March and at the HUPO 5th Annual World Congress in September.

action databases that now exchange data, as a case study. In hindsight, an index would have been a useful feature when the consortium participants worked out the bugs in the exchange mechanism, he says. Once the index is in place, each repository will start sharing newly deposited information with the other members of the collaboration. “We’ll probably first start with exchanging data on the processed peak list level,” says Apweiler. But he adds that repositories may never move raw data around, because the files are too large. In these cases, the index will simply point users to the appropriate repository. In the meantime, the ProteomEx collaborators are preparing a grant proposal to help fund their efforts. The group plans to meet again at the spring meeting of the HUPO Proteomics Standards Initiative.