Finding Clinically Relevant SNPs 

The Challenge:

Tens of thousands of people have subscribed to Direct-to-Consumer genetic testing services and it is interesting to know how that data might be useful for clinical research. SNPWeb uses SNP definitions from PharmGKB and integrates data from Affymetix Full Genome 6.0 annotations as well as sample files from 23andMe. The applet allows you to search for SNPs and see whether they are be available in 23andMe, or Navigenics which uses the Affymetrix array. 

The Application:

This application was inspired by curiosity over what SNPs are available through DTC vendors and meant to extend an analysis by Eric J. Peters. It is in prototype form now and will be extended as better data is available for clinically relevant SNPs.

The Technology:

This is a first prototype with Microsoft's Rich Internet Application (RIA) services beta code. The application defines a "middle-tier" domain which provides a structured layer between the SQL database and Silverlight presentation layer. The application is hosted in the Microsoft Azure cloud and database is hosted in SQL Azure.

Adverse Drug Reaction Prediction

The Challenge:

The rise of polypharmacy has made it difficult for clinicans to predict when combinations of drugs will have adverse effects. In addition, there is a large variation among patients in the way that they metabolize drugs due to genetic factors and common foods or health factors can complicate the dosing decision.

The Application:

GeneMedRx provides a proprietary database and an analytical engine that enables improved predictivity for adverse drug reactions. The analysis takes into account the cumulative effects of multiple drugs as well as the genotype of the patient and other foods and health factors. The application provides a more comprehensive view of drug interactions than simple pair-wise comparison of drugs.

The Technology:

This prototype was written with Silverlight 3 front end implemented over web services that interface with MySQL static database, SQL Server patient database, and the .Net analytical engine. The application is designed to integrate through the server layer to other Electronic Medical Record systems.

Genetically Informed Medication Selection 

The Challenge:

Pharmicogenetic tests are becoming common, but the interpretation of those tests requires highly specialized knowledge. To use results wisely, a clinician must understand the limitations of the test results and the relationship between a patient's genotype and their phenotype. Once that is understood, they must choose an appropriate medication based upon the patient's ability to metabolize the medication.  

The Application:

The report module accepts genotypes from either PCR (Luminex) or array-based (Autogenomics) assays. Based upon the equipment type, the application predicts the patient's most likely phenotype and provides a report that provides dosage guidelines for the predicted phenotype that are taken from the most current scientific literature. When the patient has variations in multiple genes, the GeneMedRx algorithm is used to predict the cumulative effect of the patient's genotype on their ability to process medications.

The Technology:

This application was written with a Silverlight 3 front end implemented over a collection of web services that implement data access and analysis. Reports are created using SQL Server Reporting services and delivered as PDF files.