Secure genome crowdsourcing for large-scale association studies

Abstract

Computationally-efficient techniques facilitate secure crowdsourcing of genomic and phenotypic data, e.g., for large-scale association studies. In one embodiment, a method begins by receiving, via a secret sharing protocol, genomic and phenotypic data of individual study participants. Another data set, comprising results of pre-computation over random number data, e.g., mutually independent and uniformly-distributed random numbers and results of calculations over those random numbers, is also received via secret sharing. A secure computation then is executed against the secretly-shared genomic and phenotypic data, using the secretly-shared results of the pre-computation over random number data, to generate a set of genome-wide association study (GWAS) statistics. For increased computational efficiency, at least a part of the computation is executed over dimensionality-reduced genomic data. The resulting GWAS statistics are then used to identify genetic variants that are statistically-correlated with a phenotype of interest.

1 . A method executed at a first computing entity, comprising:
 receiving a share of a first data set, the first data set comprising sequenced genomes and phenotypes of study participants, wherein the share of the first data set is received at the first computing entity in association with distribution, according to a secret sharing protocol, of shares of the first data set to a set of computing entities that include the first computing entity and at least a second computing entity;   receiving a share associated with a second data set, the second data set comprising mutually independent and uniformly-distributed random numbers and results of calculations over the random numbers, wherein the share of the second data set is received at the first computing entity in association with distribution, according to the secret sharing protocol, of shares of the second data set to the set of computing entities; and   using the share associated with the second data set to facilitate a secure and efficient computation over the share of the first data set to generate an output, the computation being carried out during an interaction between the first computing entity and the second computing entity, wherein part of the computation is executed over a compression of the share of the first data set for increased computational efficiency.   
     
     
         2 . The method as described in  claim 1  wherein the output comprises a set of genome-wide association study (GWAS) statistics. 
     
     
         3 . The method as described in  claim 2  further including using the GWAS statistics to identify genetic variants that are statistically-correlated with a phenotype of interest. 
     
     
         4 . The method as described in  claim 1  wherein the compression of the share of the first data set is generated by applying a random projection to the first data set. 
     
     
         5 . The method as described in  claim 1  wherein the second data set and the share associated with the second data set are pre-computed. 
     
     
         6 . The method as described in  claim 5  wherein the second data set is generated according to a modified Beaver triple computation that facilitates arithmetic computation over the first data set beyond pairwise multiplication. 
     
     
         7 . The method as described in  claim 1  wherein the share associated with the second data set is pre-computed by a computing entity that is distinct from the first computing entity and the second computing entity. 
     
     
         8 . The method as described in  claim 1  wherein neither the first computing entity nor any other computing entity can reconstruct the first data set from the share, thereby preserving privacy of each sequenced genome and phenotype. 
     
     
         9 . The method as described in  claim 1  wherein the first computing entity and the second computing entity are independent computing entities. 
     
     
         10 . A system for crowdsourcing genomic data for large-scale association studies, whereby privacy of individual genomic and phenotypic data is preserved, comprising:
 a computing entity comprising a hardware processor, and computer memory holding computer program instructions executed by the hardware processor and configured to:
 receive a secret share of a first data set, the first data set comprising sequenced genomes and phenotypes of different individuals; 
 receive a secret share associated with a second data set, the second data set comprising mutually independent and uniformly-distributed random numbers and results of calculations over the random numbers; and 
 use the secret share associated with the second data set to facilitate a secure and efficient computation over the share of the first data set to generate an output, the computation being carried out during an interaction between the computing entity and the second computing entity, wherein part of the computation is executed over a compression of the share of the first data set for increased computational efficiency. 
   
     
     
         11 . The system as described in  claim 10  wherein the output comprises a set of genome-wide association study (GWAS) statistics. 
     
     
         12 . The system as described in  claim 11  wherein the computer program instructions are further configured to use the GWAS statistics to identify genetic variants that are statistically-correlated with a phenotype of interest. 
     
     
         13 . The system as described in  claim 12  wherein the compression of the share of the first data set is generated by the computer program instructions configured to apply a random projection to the first data set. 
     
     
         14 . The system as described in  claim 10  wherein the share associated with the second data set is pre-computed. 
     
     
         15 . The system as described in  claim 14  wherein the share associated with the second data set is generated according to a modified Beaver triple computation that facilitates arithmetic computation over the first data set beyond pairwise multiplication. 
     
     
         16 . The system as described in  claim 10  wherein the share associated with the second data set is pre-computed by a computing entity that is distinct from the computing entity and the second computing entity. 
     
     
         17 . The system as described in  claim 10  wherein neither the first computing entity nor the second computing entity can reconstruct the first data set from the shares, thereby preserving privacy of each sequenced genome and phenotype. 
     
     
         18 . The system as described in  claim 10  wherein the first computing entity and the second computing entity are independent computing entities. 
     
     
         19 . A method for securely crowdsourcing genomic and phenotypic data for large-scale association studies, comprising:
 receiving, via secret sharing, genomic and phenotypic data of individual study participants in a manner that preserves privacy of individual genomic and phenotypic data;   receiving, via secret sharing, results of pre-computation over random number data; and   executing a secure computation against the secretly-shared genomic and phenotypic data, using the secretly-shared results of the pre-computation over random number data, wherein at least a part of the computation is executed over dimensionality-reduced genomic data for increased computational efficiency, to generate a set of genome-wide association study (GWAS) statistics.   
     
     
         20 . The method as described in claim  21  further including using the GWAS statistics to identify genetic variants that are statistically-correlated with a phenotype of interest.