Nicolas F. Fernandez¹, Andrew D. Rouillard¹, Klarisa Rikova², Peter V. Hornbeck², Avi Ma’ayan¹

1Department of Pharmacology and Systems Therapeutics, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, Box 1215, New York, NY 10029 USA
²Cell Signaling Technology Inc., 3 Trask Lane, Danvers, MA 01923 USA

Kinases are a class of cell signaling proteins that control diverse cellular functions through protein phosphorylation. Dysregulation of kinase activity is common in many cancers, and kinases are effective therapeutic drug targets. However, we still have a very partial view of the human kinome in normal physiology and disease. To address this challenge we developed a web-based tool and database that can be used to predict kinase activity given proteomics, phosphoproteomics and genomic data. KEA2 is a web based kinase enrichment analysis and network visualization tool that significantly expands on our previous published popular tool KEA. With KEA2 users can perform enrichment analyses given protein lists or list of differentially phosphorylated phosphosites. For KEA2 we assembled data from a diverse set of over twenty resources to generate different views of the human kinome. These views connect kinases based on their known binding partners, known substrates, co-expression, effects on cancer cell-lines when knocked down, and similar roles in disease. Using these resources, we generated kinase similarity networks as well as a web visualization tool for exploring kinase clusters and discovering new kinase-kinase associations across different resources. Additionally, we applied KEA2 to the analysis of original unpublished phosphoproteomics dataset collected from 31 non-small cell lung cancer cell lines. The analysis generated unique kinase signatures for the cell lines with agreement with previous knowledge as well as point to potential new drivers in lung cancer. In conclusion, KEA2 will be a useful tool to advance our understanding of the human kinome.