IBI5071-1 - Theoretical and practical metagenomic approaches to virus detection

• This course aims to present fundamental concepts of profile HMM construction using viral sequences and their application on genomic and metagenomic data. It is also an objective of the course to introduce different machine learning approaches applied to viral sequence studies.

The advent of next-generation sequencing has brought the possibility of sequencing not only a single genome but the genomes of a whole community of microorganisms of a biome. Metagenomics allows estimating the biological diversity of a sample, the whole set of enzymes and pathways present in the community, as well as to detect unknown organisms. We currently know only a small fraction of the viral diversity. The use of metagenomic data and the identification of emerging viruses represent a major challenge in terms of bioinformatics. In this discipline we intend to cover some methods and tools for processing metagenomic data and their use for the viral discovery, including theoretical concepts and practical sessions.

The advent of next-generation sequencing has brought the possibility of sequencing not only a single genome but the genomes of a whole community of microorganisms of a biome. We currently know only a small fraction of the viral diversity. The use of metagenomic data and the identification of emerging viruses represent a major challenge in terms of bioinformatics. First, viruses evolve much faster than prokaryotes and eukaryotes, leading to a higher divergence of the sequences and making their detection by conventional pairwise alignment methods more difficult. Second, the number of viral genomes available on public databases is relatively low, compared to archaea and bacteria, for instance. This aspect also makes viral sequence detection and classification much more challenging. In this course, we intend to cover some innovative methods that have been recently developed and that can increase the sensitivity of detection of evolutionary remote viruses. One of the approaches involves the construction and application of profile HMMs. We will teach conceptual aspects of profile HMM construction, especially for taxonomically specific groups of viruses. Also, we will offer practical sessions where the students will be able to build and apply profile HMMs in metagenomic data for viral detection and discovery. We also intend to cover the fundamentals of different machine learning approaches and present in practical sessions different methods applied to viral detection, classification, virus-host interactions, among other topics.

1. Metagenomics - challenges for viral detection and discovery
2. Profile HMM construction for viral detection discovery
3. Screening metagenomic data with profile HMMs
4. Targeted progressive assembly using profile HMMs as seeds
5. Finding proviruses in bacterial genomes using profile HMMs
6. Introduction to machine learning methods
7. SVM to detect viral miRNAs
8. Random Forest to detect viral miRNA
9. PCA for viral host classification
10. >CNN for viral host classification
11. Introduction into RNA world
12. Folding algorithm: MacCaskill and Partition functions
13. RNAfold to determine the secondary structures of RNA viruses
14. Vienna RNA Package for the study of virus host interactions>
15. LRIscan/Circos (long-range Interactions of segmented viruses)
16. Covariance models
17. Infernal to detect viral elements from (meta-)genomic samples

• Theory test at the end of the course.