Deep Data Analysis of a Large Microarray Collection for Leukemia Biomarker Identification

摘要

Nowadays, statistical design of experiments allows for planning of complex studies while maintaining control over technical bias. In this study, the equal importance of performing tailored preprocessing, such as batch effect adjustment and adaptive signal filtration, is demonstrated in order to enhance quality of the results. This approach is assessed on a large set of data on acute and chronic leukemia cases. It is shown, both through statistical analysis and literature research, that drawing attention toward data preprocessing is worthwhile, as it produces meaningful original biological conclusions. Specifically in this case, it entailed the revealing of four candidate leukemia biomarkers for further investigation of their significance.