In 2011, we published the Urinary Protein Biomarker Database (UPB database), a manually curated database compiling results of urinary protein biomarker studies from published literature of proteomic studies as well as small-scale experiments such as ELISA and Western blot. Manually, curation ensures the minimum mistakes appear in the process of database establishment.

All of the protein/peptides that were reported to have abundance change under disease conditions are considered as biomarker candidates and collected in this database. No extra filtration of biomarker confidence is used since the aim of building this database was to preserve the original result of literature and make the database as comprehensive as possible. Users can use information such as detection method and sample size displayed in the database to help assess data confidence themselves. Table 19.2 shows data statistics of this database.

Particularly, a very small proportion of records in this database are ‘negative records,’ in which changes of protein abundance were reported as not statistically significant under disease conditions. These records are included because the same proteins were identified as biomarker candidates for the same diseases by other studies. Including negative data in the biomarker database is important, since it may help researchers to assess data confidence better by analyzing conflict results for the same biomarker candidate.

For each biomarker candidate in this database, the following information was collected (if available): definition and sample size of the disease and control groups, experimental procedures and instrument types, protein information such as fold change, fragment, variant and post-translational modification (PTM), experimental molecular weight and pI. In particular, proteins were queried in the plasma proteome list [5] to infer its origin (Fig. 19.1).