ABSTRACT: RESEARCH PROJECT 1 Cancers of Unknown Primary (CUPs) are metastasized tumors without an identifiable primary site. They are treated with general platinum/taxane based therapy and have poor survival. We propose to identify the primary site for CUPs, based on tissue specific methylation profiles. The long-term goal is to use the primary site identity as an additional piece of information that will guide therapy. We hypothesize that there are methylation markers that are tissue specific and stable across normal and different stages of cancer cells. For Aim 1, we use discriminant analysis on methylation data from the Cancer Genome Atlas (TCGA) and others to identify the markers that tell us the tissue of origins. Using these markers, for Aim 2, we will develop a bioinformatics pipeline that can infer the primary site from the integration of DNA methylation, mRNA and miRNA markers. For Aim 3, we will validate our methods using archived samples of CUPs and other metastasized tumors by blindly testing the predictions of our algorithm. The approach is innovative, since it will be the first study to integrate gene expression and DNA methylation to predict the primary site, while also taking into account the heterogeneous cell types in an unpurified sample. The outcomes from this work will be a list of markers that are stable for tissue identification across different cancer stages, and an open-source bioinformatics software that can be used to identify primary sites of CUPs. The proposed research will increase the information that can be used to guide more personalized therapies for CUPs.