Automated quantitative assessment of proteins' biological function in protein knowledge bases

Abstract

Primary protein sequence data are archived in databases together with information regarding corresponding biological functions. In this respect, UniProt/Swiss-Prot is currently the most comprehensive collection and it is routinely cross-examined when trying to unravel the biological role of hypothetical proteins. Bioscientists frequently extract single entries and further evaluate those on a subjective basis. In lieu of a standardized procedure for scoring the existing knowledge regarding individual proteins, we here report about a computer-assisted method, which we applied to score the present knowledge about any given Swiss-Prot entry. Applying this quantitative score allows the comparison of proteins with respect to their sequence yet highlights the comprehension of functional data. pfs analysis may be also applied for quality control of individual entries or for database management in order to rank entry listings.

Figure 1

Computation of pfs. (A) pfs was calculated with regard to the validity of bibliographic citations as well as a description and comments section. In line 9, the FUNCTION and CATALYTIC ACTIVITY records were evaluated. In step 2, automated entries resulting from large-scale experimental approaches were detached from further analysis. (B) Section evaluation. In every section, every single sentence was evaluated independently. Lines 12–14 are required in case literature citations contain down-weighting phrases, yet conclusions regarding functional properties of a protein have eventually been made by mere resemblance at the primary sequence level (Swiss-Prot term: “BY SIMILARITY”).