A Novel Statistical Method to Identify Robust Prostate Cancer Marker Genes

The most commonly used diagnostic screen for prostate cancer (measurement of levels of prostate specific antigen) is now known to function very poorly as a disease biomarker. The test developed in this work exhibits performance that dramatically exceeds that of PSA. The invention is the TSP (Top Scoring Pair) classifier algorithm, which is a robust method of integrating multiple sets of data from different DNA microarray studies. The system is used to find a set of markers with inverted expression levels of RNA that can differentiate cancerous versus normal patient samples. The statistical methodology central to the invention finds a set of two biomarkers that have the biggest difference in expression between the normal and diseased states (i.e. the biomarker with the largest increase and the largest decrease would yield the biggest difference). Description (Set) Proposed Use (Set) The invention can be used to find robust, accurate, sensitive, and specific gene biomarkers for disease prediction with a smaller patient sample set than other statistical methods. The need to normalize samples across lab procedures or across labs is eliminated because the TSP algorithm finds the largest change between the normal and diseased states. This method is applicable to other types of microarray data for marker gene identification.

Inventor(s): Winslow, Raimond L.

Type of Offer: Licensing

Next Patent »
« More Medical Patents

Share on      

CrowdSell Your Patent