Selected article for: "convolutional filter and information content"

Author: Jakub M Bartoszewicz; Anja Seidel; Bernhard Y Renard
Title: Interpretable detection of novel human viruses from genome sequencing data
  • Document date: 2020_1_30
  • ID: ac00tai9_10
    Snippet: Next, we propose a new approach for convolutional filter visualization using partial Shapley values to differentiate between simple nucleotide information content and the contribution of each sequence position to the final classification score. To test the biological plausibility of our models, we generate genome-wide maps of "infectious potential" and nucleotide contributions. We show that those maps can be used to visualize and detect virulence.....
    Document: Next, we propose a new approach for convolutional filter visualization using partial Shapley values to differentiate between simple nucleotide information content and the contribution of each sequence position to the final classification score. To test the biological plausibility of our models, we generate genome-wide maps of "infectious potential" and nucleotide contributions. We show that those maps can be used to visualize and detect virulence-related regions of interest (e.g. genes) in novel genomes. Finally, we analyze a recently discovered SARS-CoV-2 coronavirus, which caused a pandemic in 2020 (Wu et al., 2020) .

    Search related documents:
    Co phrase search for related documents