• Qingpeng Zhang - Illumina, San Diego, CA
  • Susannah Tringe - JGI, Walnut Creek, CA


Viruses are important drivers of ecosystem dynamics, nutrient turnover and disease. While the role of viruses in disease is well studied, their roles in ecosystems are becoming more apparent through increased metagenomic surveys. The growing amount of metatranscriptome and metagenome data present the opportunity to find new, highly divergent RNA and DNA viruses. However, identifying highly divergent viruses requires moving beyond traditional homology based search. We are developing a classifier for metagenomic and metatranscriptomic contigs that takes information from across the feature spectrum and jointly trains wide logistic regression and deep feed-forward neural network models to identify new viruses with little homology to existing viruses.


ITSxpress on Github