Full Name
Jan Felix Finke
Job Title
Postdoctoral Fellow
Hakai Institute & University of British Columbia
City (Work Address)
State/Province/County (Work Address)
Speaker Bio
Jan is an oceanographer with a background in biology, using data science to understand the interactions among viruses, their hosts, and physical and chemical
variables in the marine environment. After completing a BSc in applied biology in Bonn, Germany and a MSc in Oceanography in Amsterdam, The Netherlands Jan moved to
Vancouver for his PhD in Oceanography. Jan has experience in a wide range of research such as biotechnology, ship ballast water treatment, aquaculture and
especially marine viral ecology. Jan’s PhD research was focused on the diversity and ecology of phytoplankton viruses. Recent research was on viral pathogens of shellfish in
the wild and in aquaculture facilities. Currently Jan is a Hakai Postdoctoral Fellow at UBC studying the impact of viruses on the marine food web. Jan is interested in
applying computational approaches and statistical models to discover and monitor viruses in the environment.
Abstract Title
The combination of metatranscriptome analysis and a deep learning algorithm produces a comprehensive assessment of viral activity in marine samples
Abstract Summary
"Viruses are major agents of microbial mortality in marine systems; yet, data on the viruses that are actively replicating is lacking. One recent approach to address this problem is through metatranscriptomics, which captures replicating viruses indiscriminately and indicates their activity. However, the extraction and taxonomic assignment of viral sequences in metatranscriptomic data remains challenging.
We used a deep learning algorithm in combination with an alignment-based approach to extract viral sequences from metatranscriptomic data and assign taxonomic lineages to viruses infecting marine microbial communities coastal British Columbia waters. These viral sequences were used to assess their activity in marine communities across samples.
The combined approaches produced high-confidence viral sequences with congruent taxonomic classifications representing the major viral groups in the four realms. Dozens of known viral families were detected with putative eukaryote and prokaryote hosts. The relative abundance of viral families varied among seasons and depths, and could be related to temperature, salinity and chlorophyll. Additionally, the deep learning algorithm extracted putative viral sequences that could not be aligned to known references, indicating novel viruses.
The present work highlights the advantages and potential of analyzing metatranscriptomic data using deep-learning algorithms to study in situ infection of microbial communities by marine viruses. The result yields high confidence taxonomic assignment of actively replicating viral assemblages that map to a broad range of viral families; thus, providing a better understanding of the ecological significance of viral infection in natural systems."
Jan Felix Finke