Emmy Noether Research Group Virus in Nature and Health

Research Focus

The Deng lab uses an integrative approach combining single-cell technologies, culture-independent techniques, multi-omics, and machine learning to understand the underlying mechanisms of phage-host interactions and their impact on the human host. They apply the knowledge gained to develop novel phage-based therapeutics for i) targeting drug-resistant bacteria and ii) preventing or iii) ameliorating severity in dysbiosis associated diseases and conditions, such as inflammatory bowel disease (IBD), colorectal cancer (CRC), graft versus host disease (GVHD), allergy, and much more by manipulating the human microbiota.


To this end, they have and will develop cutting-edge methodologies for studying phage-host relationships in vitroin vivo, and in silico.

Methods developed

Bioinformatic toolbox

Despite growing interest in recent years, studying the virome remains a challenging endeavor due to several reasons, including the scarcity of viral genomic material (compared to microbial and human nucleic acid fraction), small genome sizes of viruses and their low abundance in some cases. In a microbial community only 2–5% of the total DNA is generally of viral origin. Additionally, no conserved gene regions applicable for all viral types have been identified so far. Moreover, many viruses have not been characterized yet and are not included in viral databases. These facts impede straightforward contig assemblages as well as functional annotation of viral genomes and metagenomes.
We built a human virome protein cluster (HVPC) database in order to improve and facilitate functional annotation and characterization of human viromes. This Human Viral Metagenomic Database for Diversity and Function Annotation constitutes of 12 terabases in total, more than 6 Million open reading frames (ORFs), and 927K function clusters. Further improvements such as using machine learning tools (random forest ad neural networks/deep learning) to identify host of viruses using signals including CRISPR, prophages, k-mer, etc. are currently on-going.

Figure 1. Venn diagram showing the distribution of HVPC clusters between different body sites, illustrating that ORFs from the same body site are clustering together (Elbehery et al., 2018, Frontiers in Microbiology).

Viral tagging

Viral-Tagging” (VT) is a high-throughput, culture-independent means of experimentally linking wild viruses to a target host, and vice versa. The DNA of uncultivated viruses is labeled non-specifically with a fluorescent dye, then viruses are mixed with a ‘bait host’, and infected cells are collected by fluorescence-activated flow cytometric sorting. The infecting viral DNA is quantitatively amplified to produce viral-tagged metagenomes. VT enables researchers to broadly map how viruses change over space and time.

Figure 2. Overview of “viral-tagging” technique (Deng et al, mBio2012; Nature 2014; Dzunkova et al, Nature Microbiology, 2019).