SMART Genomics Advisor Tutorial Hello and welcome to a demonstration of the SMART Genomics Advisor.

We will demonstrate, among other things, merging a patients genomic and non-genomic clinical information into an integrated clinical informatics display. This video addresses diabetes mellitus type 2. While we can run the SMART Genomics Advisor App standalone, we will first show it integrated into a version of the SMART Diabetes Monograph (see Figure S1, S2). The standard SMART Diabetes Monograph app presents blood pressure, vitals, cholesterol, other key labs and other information germane to diabetes type 2. Sample data come from the SMART Reference system using the normal SMART application programming interface. The SMART Genomics Advisor App integration brings in patient genetic data from a secondary database here an instance of openSNP and genomics analytics. In our integrated SMART Diabetes Monograph, we present genomic information directly on the main screen, leaving a deeper dive into SNP-level detail to a pop-up screen Looking at the top right, we see the genomics risk conditions: diabetes type 1, diabetes type 2, hypertension, and coronary heart disease. These genetic disposition for having them is important to problems have not developed. graph for selected diabetes related and two serious co-morbidities, are presented because a patients clinical decision-making even if the

The risk is a numeric factor, that is, a multiplier, relative to the population norm. A risk factor of one (one-X) represents the overall normal population risk. In the graph, for example, a bar which reaches 1X for diabetes mellitus type 1, or DM1, means the patients genetic risk for being or becoming DM1 is 1, exactly the population average. A bar reaching 2X would be twice, a bar reaching 1/2X, would mean half the population average. If the risk factor is greater than 1 (blue is baseline) the section above the half way line would be colored red; if less, it is colored green. Looking lower, we see the bar heights reported numerically, colored-coded them green for low and red for high risk. The last piece of integrated genomic information is medication advisories. These represent gene-drug interactions relative to the patients current medication list: how

our patients genome affects a medications impact in terms of effectiveness, sensitivity, morbidity, or mortality. Genomic risk on the main screen is calculated from individual SNP-details in the patients genome. To see this detail that, we click the Genomics Advisor link. To the left (see Figure S2), we show a list of specific SNPs, patient genotype and associated genetic information group by each of the four conditions. Individual SNPlevel risk is used to calculate the summary genomic risk, color coded exactly as we saw on the main screen. The numbers are color-coded here, too: green for low, orange for medium, and red for high genomic risk. To the right, we show these data in one radar graph per risk condition. Each graph has as many axes as there are contributing SNPs that contribute to the aggregate risk per risk condition. The relative risk for each SNP is its distance from its graphs origin. The summary genomic risk is shown on the border of each graph. The Diabetes Type 2 graph region is larger because it is the focus of the SMART Diabetes Monograph app. On the pop-up, we add something new, namely, a disease information summary, based on risk for each of the SNPs and all conditions, including the comorbidity conditions as they are relevant for recommending treatment and/or further testing. We also repeat the medication advisories information exactly as seen on the main monograph screen. Because any disease that has known genetic factors can be captured to a good first approximation in a single SMART Genomics Advisor screen, we have also created a standalone for the SMART Genomics Advisor App (see Figure S3). Launching this app, we can see how, unlike the integrated version, the summary and detail information are consolidated into a single screen. Certain sizing tweaks are also applied to the radar graphs. The SMART API brings in relevant patient data since, as before, they can help trim results to current problems, e.g., issuing gene drug interaction advisories only for the current medication list. The genomic data come, as before, from a complementary source. Of course, we intend to enhance the SMART Genomics Advisor App in many ways, including for dynamic diseases, which have evolving genomic risk factors. For example, cancer will involve connecting the progression of the actual disease with changing gene expression. In summation, by engineering a sophisticated clinical app mash up of data, calculation, and visualization, our SMART Genomics Advisor-augmented SMART Diabetes Monograph App incorporates genomics for everyday clinical care of diabetes. Importantly, we hope our demonstration hints at how this informatics approach can be

generalized for many diseases. Our demonstration of our standalone SMART Genomics Advisor App is but a first step toward more complex diseases and broader omics integration. We welcome comments and questions at genomics@smartplatforms.org

Figure S1. SMART Diabetes Monograph main screen and integrated Genomics Advisor summary

Figure S2. SMART Diabetes Monograph integrated Genomics Advisor SNP-detail overlay

Figure S3. Standalone SMART Genomics Advisor App