Cancer is a systemic disease involving dysregulated biological processes of cell proliferation, metabolism, and apoptosis. It is known that some types of cancer have longer life span, and they are even curable if they are diagnosed and treated properly in the early stage. So it is essential to find biomarkers to detect these cancers in their early stages. With the rapid development of high-throughput microarray and sequencing technologies, many biomarker-based cancer early diagnosis assays are proposed and some are already available in the market. Most of the cancer biomarkers are detected through comparing cancer samples versus normal samples in a certain cancer type, but most of them are not in the comparison against other cancer types. In this research, we propose a novel computational method to comprehensively detect highly accurate cancer biomarkers for different groups of cancer types, with a special emphasis on the detection specificity against the control samples including both those from healthy persons and those from other cancer types. Such biomarkers are called specific biomarkers for a given cancer group, which may be defined as cancers of the same type, cancers with similar survival rates, grade, development stage, or cancers in the same human body systems, etc. The proposed algorithm is extensively evaluated across eight cancer types, and the detection performance shows that the specific biomarkers have reasonable sensitivities and very high specificities. The main contributions of this work are (a) the detection of highly specific biomarkers for eight cancer types and (b) the detection of specific biomarkers for cancers with the similar survival rates. The proposed algorithm may also be used to detect specific biomarkers for cancers of given stages, grades or belonging systems, etc.
- specific biomarker
- microarray data
- multiple cancer types
- survival rate
Du, W., Cao, Z., Wang, Y., Zhou, F., Pang, W., Tian, Y., & Liang, Y. (2015). Specific biomarkers: detection of cancer biomarkers through high-throughput transcriptomics data. Cognitive Computation, 7(6), 652-666. https://doi.org/10.1007/s12559-015-9336-x