Pancreatic cancer is one of the most deadly diseases in the U.S. It is hard to diagnose early, and it does not respond to treatments when discovered late. Therefore, new methods for early diagnosis and prevention are critical. Currently, our approach to finding cancer biomarkers relies on technologies that lack spatial or temporal resolution for discriminating individual cells and tumor regions. In fact, much of our analyses are based on average measurements from the mixed population of different cell types within the tumor tissue. This means that each biomarker has to be validated in multiple experimental and pre-clinical settings through very time-consuming and expensive processes, severely hampering our ability to discover diagnostic or therapeutic biomarkers. We developed a novel method to image and sequence DNA and RNA genome-wide without extracting them from the tissue, and the nucleic acid sequence is visualized directly under the microscope. Therefore, we combine positional features associated with cancer progression and molecular or genetic features associated with cancer clonal evolution. Our proposal will determine genetic sequences associated with each pixel of cancer tissue images to generate a map of genetic alteration and biomarkers as a function of the tissue landscape. If successful, our proposal could signal a new approach to discovering genetic biomarkers using specific architectural hallmarks of cancer, rather than average gene expression differences between heterogeneous tissues.