TRD 2: PREDICTIVE NETWORKS – PROJECT SUMMARY A major promise of genomics is to enable precision medicine, through use of a patient's genome and physiological state to improve treatment efficacy and outcome. Routine use of genomics data in clinical research still faces major barriers, however, including some very important challenges posed by `Big Data'. First, modern genomic datasets are typically so large and complex that most biomedical researchers or clinicians have neither the computational infrastructure nor data mining expertise to cope with them. Handling even a few hundred patients requires the ability to store, access, process, and analyze petabytes (1012 bytes) of genomic data. Second, although mainstream computer scientists and information technology companies are becoming very astute at data mining, understanding big biomedical data is likely to require a depth of understanding in physiology, biotechnology, and cellular and molecular mechanism that mainstream data analysts simply do not have. Rather, encouraging preliminary results from the NRNB and several other groups indicate that an effective way to address the challenge of Big Biomedical Data is to integrate and interpret these big data sets against appropriate representations of biological, physiological, and clinical knowledge. Biological network models at multiple scales are increasingly recognized as a natural way to represent and visualize knowledge about biological mechanisms and relationships. While the field of Network Biology has focused mostly on descriptive models of network structure, there is increasing evidence that network knowledge can also be used to guide biological and clinical predictions. In some cases, these network-guided approaches have yielded predictions of higher accuracy and / or robustness with less input or training data than traditional `black-box' machine learning methods. In this TRD project, we will pursue novel methodology for using biological network information to predict the outcome of therapy in a given patient, to identify which networks and pathways are affected by mutations conferring risk of a disease, and to predict drug response and identify novel drug targets. The major deliverable is a bioinformatic framework to integrate patient molecular and clinical data with biological network information, with the goal of making clinically-relevant diagnoses and predictions about an individual based on their genomic information.

TRD 2: PREDICTIVE NETWORKS – PROJECT NARRATIVE Analysis of personalized human genomic data promises major medical advances. However, the routine use of genomics data in medical research is in its infancy, due mainly to the challenges of working with highly complex `Big Data'. Use of biological networks promises to help organize, analyze and integrate these data into models that can be used to make clinically relevant diagnoses and predictions about an individual.