GEARS: GEnomic Analysis Research with Security

We propose the GEARSGEnomic Analysis Research with Securityeffort with the broad goal and vision of our proposal is to enable collaboration and joint analysis of medical datawithout compromising data ownersrights and complying with regulation and privacy concernsThis is achieved by introducing novel technologies from the domain of advanced cryptography that enable keeping raw data encrypted even while analyzing and computing on itThis capability is critical in order to protect against onerous data use agreementscomply with HIPAA regulations and yet speed up collaborations without the risk of re identificationIn this proposalwe focus on GWAS like computations on genomic phenotypic dataWe plan to design platforms that enable researchers and commercial entities to efficiently share and compute over sensitive whole genome and whole exome data from disparate sources while it remains encrypted at all timesData providers will both delegate to computation hosts the ability to run computations on their encrypted dataandapprove which results of computation can be decrypted and shared with othersThe computation host can merge encrypted data setsfrom multiple partiesand enable the computations on the larger aggregated setOur goal is to exhibit the usefulness to advance research and treatment on common and rare diseasesand in a later stage apply our methods for match finding in the case of rare diseases that affect small percentages of the populationThe case of rare diseases and rare variant studies is especially interesting as our methods make it easiertoshareprotectedaccesstovarieddatasetsacrosstheworldThe impact of our approach is to transform research and commercial abilities to collaborate on clinical and genomic data among all parties involved in the healthcare ecosystemwhile preserving patientsprivacyThis approach will enable to accessintegrate and use larger and richer data sets quicker and more effectivelyby simplifying onerous data use agreementsreducing the level of security barriers and speeding up collaboration andinnovationtobetteraddressbothcommonandrarediseaseneedsOur research will result in a prototype software implementationThis software will encrypt genomic data and will support GWAS computations in a manner that it is transparent to users that the data underneath is encryptedWe have the exact right team to succeed in our proposed researchOur team includes MIT cryptographers who have been formulating the underlying security primitives for our offeringan experienced technical team leader who has been leading homomorphic encryption implementation efforts for DARPA since their discoveryworld class medical genomics experts and experienced entrepreneursOur team has demonstrated that HME can be feasible for clinical and or research application in several early prototypesThis early prototype work builds on extensive academic research into HMEand the open source PALISADEencryptionlibraryimplementationsbyourteamWe propose the GEARSGEnomic Analysis Research with Securityeffort with the broad goal and vision of our proposal is to enable collaboration and joint analysis of medical datawithout compromising data ownersrightsandcomplyingwithregulationandprivacyconcernsThe outcomes of our research are new techniques and their prototype implementations that provide privacy preserving data sharing and GWAS on sensitive genomic and phenotypic data joined from multiple sourcesOur approach is based on the application of homomorphic encryptionwhich when applied at scale will transform the ability to collaborate on genomic dataderived in clinical and research settingamong all parties involvedinthehealthcareecosystemwhilepreservingpatientsprivacyandcollaboratingpartiesIP