Secure secret-sharing-based crowdsourcing for large-scale association studies of genomic and phenotypic data
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 for large-scale association studies of genomic and phenotypic data shared by study participants, comprising:
at a first computing entity CP 1 :
(a) receiving an input value [x] that has been secret-shared by the study participants to each of the first computing entity CP 1 and a second computing entity CP 2 so as not to reveal to either computing entity the genomic and phenotypic data of any study participant;
(b) subtracting from the input value [x] a secret random value [r] that has been secret-shared by an auxiliary computing entity CP 0 , the auxiliary computing entity CP 0 having generated randomly chosen blinding factors [r i ], . . . [r n ] of the secret random value [r];
(c) publishing shares of [x−r] to reveal a blinded value x−r;
(d) performing a computation over at least the published shares to generate a result;
(e) receiving results of additional computations by the auxiliary computing entity CP 0 over the blinding factors, the results of the additional computations having been secret-shared by the auxiliary computing entity CP 0 and comprising uniformly random values;
(f) using the result of the computation together with the results of the additional computations to generate a share;
(g) receiving a corresponding share from the second computing entity CP 2 , the corresponding share having been generated at the second computing entity CP 2 according to steps (a)-(f) with respect to information independently secret-shared to the second computing entity CP 2 by the study participants and the auxiliary computing entity CP 0 ; and
(h) using the share and the corresponding share received from the second computing entity CP 2 to construct an output.
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 identifying genetic variants that are statistically-correlated with a phenotype of interest using the GWAS statistics.
4. The method as described in claim 1 wherein the results of additional computations are computed by the auxiliary computing entity CP 0 in advance of steps (a) through (h) occurring at the first computing entity CP 1 .
5. The method as described in claim 4 wherein the results of additional computations are Beaver multiplication triples of correlated random values.
6. The method as described in claim 1 wherein the auxiliary computing entity CP 0 is different from the first computing entity CP 1 and from the second computing entity CP 2 .
7. The method as described in claim 1 wherein the first computing entity CP 1 cannot reconstruct the corresponding share received from the second computing entity CP 2 , thereby preserving privacy of the genomic and phenotypic data.
8. The method as described in claim 1 wherein the first computing entity CP 1 and the second computing entity CP 2 are computing entities that differ from one another.