Manuscripts

alt_text Cryptographic Aspects of DNA Screening. Preprint.

Carsten Baum1, Hongrui Cui2, Ivan Damgård1, Kevin Esvelt3, Mingyu Gao4, Dana Gretton3, Omer Paneth3,5, Ron Rivest3, Vinod Vaikuntanathan3, Daniel Wichs3,5, Andrew Yao4, Yu Yu2,4

1Aarhus University

2Shanghai Jiao Tong University

3Massachusetts Institute of Technology

4Tsinghua University

5Northeastern University

Cryptographic Aspects of DNA Screening

Abstract

Securely screening synthetic DNA orders is crucial to minimizing the number of individuals and groups capable of accessing biological weapons of mass destruction, but it must be accomplished without disclosing information on potential bioweapons. Accomplishing this goal requires: 1) screening orders against a database of hazardous sequences, whose secrecy should be protected at the highest possible level while preserving usability; 2) protecting the privacy of the client synthesizer's queries. In this document, we propose a cryptographic screening protocol that accomplishes these objectives, providing accurate complexity-theoretical assumptions, precise security guarantees, and rigorous mathematical proofs. In addition to technical details, we also explain the reasoning behind our protocol design for the benefit of non-cryptographers.


alt_text Fast, accurate, and secure DNA synthesis screening with random adversarial thresholds. Preprint.

Dana Gretton1, Erika A. DeBenedictis1, Andrew B. Liu2, Andrew C. Yao3,*, Kevin M. Esvelt1,*

1Massachusetts Institute of Technology

2Harvard Medical School

3Tsinghua University

*Corresponding authors

Fast, accurate, and secure DNA synthesis screening with random adversarial thresholds

Abstract

Global DNA synthesis is growing faster than Moore’s Law, but not all DNA synthesis machines screen for sequences that could be used to build biological weapons of mass destruction. Current approaches are computationally intensive, cannot be automated due to high false positive rates, and unavoidably disclose which sequences are considered threats. Here we describe a novel architecture that looks for exact matches within a database of randomly selected fragments from possible bioweapons as well as potentially functional variants. Removing database entries matching unrelated sequences in GenBank can eliminate nonrandom false positives, enabling automation using algorithms permitting oblivious encryption. Random adversarial threshold screening may offer a fast, automated, and secure method of safeguarding the international community from synthetic bioweapons of mass destruction.

中文 English