Secure Algorithms for Biomedical Research in Public Clouds

Abstract

Algorithms from the biomedical domain have to face a rapid growth of biological data and therefore a rising demand for computing time. The predictive power of such algorithms is also further improving and becomes increasingly interesting for commercial applications. Cloud Computing – as an already established paradigm to elastically allocate computing resources on demand – offers flexible solutions to deal with the increasing request for compute power. However, security concerns remain when valuable research or business data are being processed in a Public Cloud. Herein, we describe – from the application and security perspective – three biomedical case studies from different domains: Patent annotation, cancer outcome prediction, and drug target prediction developed within the GeneCloud project. Our approach is to realize a data-centric security method to be able to compute on encrypted or blinded data in any non-trustworthy environment accessible by the user.