Biomedical data sharing with privacy

Professor Bonnie Berger is the Simons Professor of Mathematics at MIT, holds a joint appointment in Electrical Engineering and Computer Science, and serves as head of Computation and Biology group at MIT's Computer Science and AI Lab. Her recent work focuses on designing algorithms to gain biological insights from advances in automated data collection and the subsequent large data sets drawn from them. She works on a diverse set of problems, including Compressive Genomics, Network Inference, Structural Bioinformatics, Genomic Privacy, and Medical Genomics. Additionally, she collaborates closely with biologists in order to design experiments to maximally leverage the power of computation for biological explorations.

Volterra lecture 

Time: 13:00 - 14:00 Friday 9 November 2018

Place: Room 3.069, Chemistry Building 3, NTNU, Sem Sælands vei 6-8, Trondheim

The event is co-organized with the ERASysAPP project 

Abstract:

The last two decades have seen an exponential increase in genomic and biomedical data. Extracting new science from these massive datasets will require efficient cryptographic techniques that can help overcome the privacy barriers to biomedical data sharing. Leveraging and adapting tools from modern cryptography, we present the first scalable, end-to-end pipeline for secure genome-wide association studies (GWAS), which obtains accurate results while ensuring no information about the underlying data is leaked to any entity, including the researchers. Our approach enables secure genome crowdsourcing at the scale of a million genomes, allowing the individuals to contribute their genomes to a study without compromising their privacy.

Also, we introduce a protocol for securely training a neural network model of drug-target interaction (DTI) that newly ensures the confidentiality of all underlying drugs, targets, and observed interactions. Our protocol scales to a real dataset of more than a million interactions, and is more accurate than state-of-the-art (plaintext) DTI prediction methods. Using our protocol, we discover novel DTIs that we experimentally validated via targeted assays. Our work lays a foundation for more effective and cooperative biomedical research.


Information about the event

  • Time:
    09. — 09. Nov 2018
  • Place:
    NTNU
  • Event type:
    Volterra Lectures

Contact

Gunnar Dick

gunnar.dick@mn.uio.no
404 54 144

Profile

#Volterra Lectures #Helse