Analysis of antibody microarrays high throughput data: from data processing to network inference. The case of B-cell chronic lymphocytic leukemia.

A particular category of protein arrays, namely antibody microarrays, is becoming increasingly important in proteomic research. Consistent developments in the least few years led to the possibility to detect expression and phosphorylation level of hundreds of proteins at once in biological samples with unprecedented sensitivity, still unmatched by other methods, such as mass-spec. Being a relatively new technique, and considering the wealth of information potentially generated, the antibody microarrays approach requires ad-hoc strategies in terms of signal quantification and filtering, data normalization, statistical analysis and functional interpretation. In this study, several aspects are discussed, with particular focus on the antibody microarrays technology developed by the company Kinexus, CA USA.A meaningful example of the application of the discussed criteria is presented, in the context of a large study of human leukemia samples we performed, including 34 B-cell chronic lymphocytic leukemia and 10 healthy human subjects before and after stimulation with the chemokine CXCL12, analyzed by means of 44 arrays monitoring over 800 antibodies, also including about 300 antibodies detecting phosphosites. Comparison strategies and network analysis approaches are described. We used the network analysis platform Cytoscape (http://www.cytoscape.org/) along with plugins developed by our group (PesCa, CentiScaPe, www.cbcm.it) to highlight key pathways suggested by array data. Particular emphasis is given to the impact of microarray data on proteins network reconstruction and understanding of fundamental biological processes.