Mass spectrometry information has long offered the potential of discovering biomarkers that would enable clinicians to diagnose disease, and treat it with targeted therapies. Hundreds of human samples alone have been used to generate thousands of spectra for identification. This data, and the generation of targeted peptide information, represents the first step in the process of locating disease biomarkers. Reaching the goal of clinical proteomics requires that this data be integrated with additional information from disease literature and genomic studies. Here we describe PeptideAtlas and associated methods for mining the data, as well as the software tools necessary to support large-scale integration and mining.