“…What is apparent is that there is a need for studies to test whether such an approach is feasible, potentially informative, and capable of generating enough of a "signal" of disease association that can be discriminated from the "noise" of the multiple variables that can confound these studies. Sources of variability need to be understood: There are many potential sources of variability that can affect EWAS data interpretation, including patient or disease issues (age (Heyn et al, 2012), sex (Sarter et al, 2005), and medications (Gonzalez-Fierro et al, 2011;Junien, 2006) or exposure histories), sample collection issues, nucleic acid purification protocols (Soriano-Tarraga et al, 2013), influences inherent to the experimental assays performed, and even the version and type of software analytical tools used to process and interpret the data generated. To account for these influences, it is recommended that metadata (data describing the information collected and generated) also be collected systematically and comprehensively.…”