Integrated Genomic Surveillance reveals extensive onward transmission of travel-imported SARS-CoV-2 infections in the community (preprint)

Integration of genomic surveillance with contact tracing provides a powerful tool for the reconstruction of person-to-person pathogen transmission chains. We report two large clusters of SARS-CoV-2 cases ("Delta" clade, 110 cases combined) detected in July 2021 by Integrated Genomic Surveillance in Dusseldorf. Structured interviews and deep contact tracing demonstrated an association to a single SARS-CoV-2 infected return traveller (Cluster 1) and to return travel from Catalonia and other European countries (Cluster 2), highlighting the importance of containing travel-imported SARS-CoV-2 infections.

Characterization of SARS-CoV-2 genetic structure and infection clusters in a large German city based on integrated genomic surveillance, outbreak analysis, and contact tracing (preprint)

Viral genome sequencing can address key questions about SARS-CoV-2 evolution and viral transmission. Here, we present an integrated system of genomic surveillance in the German city of Dusseldorf, combining a) viral surveillance sequencing, b) genetically based identification of infection clusters in the population, c) analysis of hospital outbreaks, d) integration of public health authority contact tracing data, and e) a user-friendly dashboard application as a central data analysis platform. The generated surveillance sequencing data (n = 320 SARS-CoV-2 genomes) showed that the development of the local viral population structure from August to December 2020 was consistent with European trends, with the notable absence of SARS-CoV-2 variants 20I/501Y.V1/B.1.1.7 and B.1.351 until the end of the local sampling period. Against a background of local surveillance and other publicly available SARS-CoV-2 data, four putative SARS-CoV-2 outbreaks at Dusseldorf University Hospital between October and December 2020 (n = 44 viral genomes) were investigated and confirmed as clonal, contributing to the development of improved infection control and prevention measures. An analysis of the generated surveillance sequencing data with respect to infection clusters in the population based on a greedy clustering algorithm identified five candidate clusters, all of which were subsequently confirmed by the integration of public health authority contact tracing data and shown to be represent transmission settings of particular relevance (schools, care homes). A joint analysis of outbreak and surveillance data identified a potential transmission of an outbreak strain from the local population into the hospital and back; and an in-depth analysis of one population infection cluster combining genetic with contact tracing data enabled the identification of a previously unrecognized population transmission chain involving a martial arts gym. Based on these results and a real-time sequencing experiment in which we demonstrated the feasibility of achieving sample-to-turnaround times of <30 hours with the Oxford Nanopore technology, we discuss the potential benefits of routine ultra-fast sequencing of all detected infections for contact tracing, infection cluster detection, and, ultimately, improved management of the SARS-CoV-2 pandemic.

Swarm Learning as a privacy-preserving machine learning approach for disease classification (preprint)

Identification of patients with life-threatening diseases including leukemias or infections such as tuberculosis and COVID-19 is an important goal of precision medicine. We recently illustrated that leukemia patients are identified by machine learning (ML) based on their blood transcriptomes. However, there is an increasing divide between what is technically possible and what is allowed because of privacy legislation. To facilitate integration of any omics data from any data owner world-wide without violating privacy laws, we here introduce Swarm Learning (SL), a decentralized machine learning approach uniting edge computing, blockchain-based peer-to-peer networking and coordination as well as privacy protection without the need for a central coordinator thereby going beyond federated learning. Using more than 14,000 blood transcriptomes derived from over 100 individual studies with non-uniform distribution of cases and controls and significant study biases, we illustrate the feasibility of SL to develop disease classifiers based on distributed data for COVID-19, tuberculosis or leukemias that outperform those developed at individual sites. Still, SL completely protects local privacy regulations by design. We propose this approach to noticeably accelerate the introduction of precision medicine.