Publication Synopsis

Integrative Genomic Data Mining for Discovery of Potential Blood-Borne Biomarkers for Early Diagnosis of Cancer

"The IPA biomarker module enables researchers to 'fish out' biologically meaningful signatures from a variety of entities including genes, miRNAs and proteins. Moreover, the biomarker comparisons feature has allowed us to predict common or unique markers across multiple datasets and/or under different disease conditions."

— Dr. Yongliang Yang, Research Fellow at Harvard Medical School, forthcoming Assistant Professor at the Center of Molecular Pharmacology, Dalian University of Technology in China

This month we are featuring a study led by a team of researchers from the Department of Radiology at Harvard Medical School that aimed to identify blood-borne biomarkers of human cancer that could be further tested and validated for use in the clinic. By mining the gene expression data stored in the Oncomine database and analyzing those candidate markers with IPA-Biomarker™ they were able to identify markers common across 6 tumor types (prostate, breast, lung, colon, ovary, and pancreas) as well as subsets that were unique to certain tumors.

The Oncomine database enabled the team to focus on genes that encode secreted proteins that are overexpressed in cancer. Analysis of those genes in IPA prioritized the candidates that met key biological criteria (human proteins detected in blood, plasma, other easily accessible biofluids), and that had evidence implicating them in cancer and cancer-related processes. IPA-Biomarker also enabled list comparisons that identified subsets of markers common or unique to each tumor type or disease state (benign and malignant), ultimately resulting in the identification of MMP-1, CD44, CP, and NOTCH4 as potential markers for breast, colorectal, ovarian, and pancreatic cancer respectively. This study establishes a biologist-friendly in silico approach to take full advantage of existing biomedical data for identification and prioritization of potential diagnostic markers for cancer. Importantly, this workflow can be adapted to mine additional databases and support other steps in the drug discovery process, including target identification.