SPD--a web-based secreted protein database

With the improved secreted protein prediction approach and comprehensive data sources, includ-ing Swiss-Prot, TrEMBL, RefSeq, Ensembl and CBI-Gene, we have constructed secretomes of human, mouseand rat,with a total of 18 152secretedproteins. All the entries are ranked according to the prediction confidence. They were further annotated via a pro-teome annotation pipeline that we developed. We also set up a secreted protein classification pipeline and classified our predicted secreted proteins into different functional categories. To make the dataset more convincing and comprehensive, nine reference datasets are also integrated, such as the secreted proteins from the Gene Ontology Annotation (GOA) system at the European Bioinformatics Institute, and the vertebrate secreted proteins from Swiss-Prot. All these entries were grouped via a TribeMCL based clustering pipeline. We have constructed a web-based secreted protein database, which has been publicly available a