In October 2019, we attended the iDASH workshop in Bloomington, IN at the Indiana University Luddy School of Informatics, Computing, and Engineering. iDASH stands for Integrating Data for Analysis, Anonymization, and Sharing, and each year (we believe since 2014) they organize a ‘Secure genome analysis competition’. We are very competitive (especially Gamze) so naturally we wanted to get involved.
The iDASH organizers set four tracks of the competition, each a different challenge. The 2019 challenges were:
Track I: Distributed Gene-Drug Interaction Data Sharing based on Blockchain and Smart Contracts
Track II: Secure Genotype Imputation using Homomorphic Encryption
Track III: Privacy-preserving Machine Learning as a Service on SGX
Track IV: Secure Collaborative Training of Machine Learning Model
These challenges were exciting to us because they each took place at the cutting edge of privacy research. The idea of storing gene-drug interaction data in the Ethereum blockchain is positively odd–but many of the best ideas are. They were also learning opportunities for us; when we got started, neither of us was an expert in the methods mentioned in the tasks. This meant we were going to learn a lot in the process of designing solutions. The competition was a way for a community of scientists and technical specialists to benchmark methods to protect individual privacy and data security in the context of genomic data.
Challenges were announced in April, we designed and implemented our solutions over the summer, and the competition culminated in a workshop on October 26 in Bloomington, IN.
At the workshop our team, ‘Team Gerstein Lab,’ was awarded 3rd place for the Track 1 challenge. We gave a talk describing our approach to storing gene-drug interaction data in a smart contract. This was Charlotte’s first time giving a talk at any sort of conference or workshop!
Aside from the exciting technical advances presented in the workshop, one of our favorite things about the day was the number of women speakers we saw (including us). In February 2018 Gamze attended a workshop on genomic privacy in Lausanne, Switzerland. While it was a great event, she was frustrated that among the many speakers there was only one woman. She complained through a tweet and as a response organizers apologized and said “sorry, there are not many female experts in the field of genome privacy that we could invite”. This year’s iDASH workshop revealed that this is not the case… there are indeed many women experts in this field as evidenced from not only the number of female invited speakers but also the workshop attendees and challenge participants.
Throughout the day there were several talks, between three and five per track. Dr. Li Xiong, Professor of Computer Science at Emory University, gave a fascinating keynote talk on the use of differential privacy for health data analysis. Then, the winning teams within each track gave talks. In Track I (which included our talk), the main challenge people faced was engineering– how to actually configure the Ethereum blockchain network and deploy your solution to the chain. The other tracks faced their own issues, for example with the SGX challenge, how to work within the memory constraints imposed by the SGX enclave size.
The day wrapped up with a panel discussion with Xiaofeng Wang (IU; Moderator), Anamaria Costache (Intel), Xiang Xie (PlatON), Rundong Zhou (Baidu), Mariya Georgieva (Inpher), Gamze Gursoy (Yale). This was really useful; it created space for experts to informally discuss some of the more nuanced issues in the field. For example, how can we develop tools that scientists and bioinformaticians will actually use? Or how can we trust privacy-protecting tools developed by for-profit companies, such as Intel? One challenge that was mentioned was that many of the bioinformaticians who work with private data are not even aware that much of the computations can now be done in the encrypted space.
Genome privacy and security are relatively new areas of study and concern, and we loved joining the iDASH community to learn more about them and contribute our solutions and ideas. Already looking forward to iDASH 2020!
iDASH 2019 website can be found at http://www.humangenomeprivacy.org/2019/index.html