Report

Deriving genomic diagnoses without revealing patient genomes

See allHide authors and affiliations

Science  18 Aug 2017:
Vol. 357, Issue 6352, pp. 692-695
DOI: 10.1126/science.aam9710
  • Table 1 Summary of results for different secure genomic multiparty computation scenarios, all using real patient data.

    N/R, not relevant.

    OperationRelevant information for each operationRunning time measurements
    MAX
    (over genes)
    Scenario:
    small
    disease
    cohort
    No. of
    unrelated
    probands
    (who avoid
    openly sharing
    their data)
    No. of rare
    functional
    variants
    (genes) per
    proband
    (median)
    No. of
    probands
    with rare
    functional
    variant/s in
    gene (top
    three,
    descending
    order)
    Gene nameProven
    causal gene
    for disease
    Protection
    quotient
    (1 – no. of
    variants shared of top gene/total no. of variants)
    Bandwidth
    (gigabytes)
    Compute (s)Network (s)
    Freeman-
    Sheldon
    syndrome
    3258 (253)3MYH3MYH31 – 3/767 = 99.6%0.020.154.91
    2DBT
    1ACADVL
    Hajdu-
    Cheney
    syndrome
    7278 (272)6NOTCH2NOTCH21 – 6/1853 = 99.7%0.030.187.29
    3HLA-DRBI
    3MCC
    Kabuki
    syndrome
    10262 (257)8KMT2DKMT2D1 – 8/2754 = 99.7%0.040.229.59
    3COL6A1
    3FLNB
    Miller
    syndrome
    4267 (258)4DHODHDHODH1 – 8/1063 = 99.2%0.030.187.29
    3DNAH5
    2ACOX2
    SETDIFF
    (over variants)
    Scenario:
    familial
    Family memberNo. of rare
    functional
    variants
    (not shared)
    No. of
    proband-only
    variants
    (revealed)
    Gene nameProven
    causal gene
    Protection
    quotient
    Bandwidth
    (gigabytes)
    Compute
    (min)
    Network
    (min)
    TrioFather185N/RN/RACTB1 – 2/524 = 99.6%18.11.756.7
    Mother164N/RN/R
    Proband1752ACTB
    USH2A
    INTERSECTION
    (over variants)
    Scenario:
    two hospitals
    No. of
    suspicious
    variants
    (not shared)
    Total intersecting variants (for patient phenotype comparison follow-up)Protection quotientBandwidth
    (gigabytes)
    Compute
    (min)
    Network
    (min)
    Washington57341591 – 318/11,067 = 97.1%3.10.379.4
    Baylor5333

Supplementary Materials

  • Deriving genomic diagnoses without revealing patient genomes

    Karthik A. Jagadeesh, David J. Wu, Johannes A. Birgmeier, Dan Boneh, Gill Bejerano

    Materials/Methods, Supplementary Text, Tables, Figures, and/or References

    Download Supplement
    • Materials and Methods
    • Figs. S1 to S6
    • References

Stay Connected to Science

Navigate This Article