DAVID annotation of clusters in the IP lists.

All three datasets were analyzed separately. Protein lists were uploaded in DAVID and analyzed against a random set of genes as background. A p-value was calculated for all GO terms, and those GO terms that were significantly overrepresented compared to the background set were clustered. Subsequent ranking of clusters is determined by the combined p-value. For each cluster a representative GO term is shown in the table. Full tables are in S4. The pathways that relate to vesicles are enhanced in bold.