Characterizing "typical" Peptides

Parag Mallick
Institute for Systems Biology

Large proteomic datasets typically contain multiple repeat identifications of proteins. This allows the empiric definition of (idiotypic) signature peptides for identified proteins based on the frequency with which particular peptides are observed. Moreover, feature analysis of idiotypic and non-idiotypic peptides allows predicting observable peptides from proteins that have not yet been identified. A database of proteins and their idiotypic peptides can for example be used for automated assessment of database search results which proves particularly helpful in cases where the primary data is weak or otherwise ambiguous. Four large yeast datasets have been analyzed for physico-chemical properties that discriminate typical peptides from other peptides.