Tonight I continued watching the London Calling 2023 Showcase Stage recordings for sessions on targeted sequencing. The speaker was Lukas Weilguny a Ph.D. student from the EMBL-EBI in the UK. The title of the session was “Dynamic, adaptive sampling during nanopore sequencing using Bayesian experimental design.” They worked on a model that takes the sequencing data and derives scores about the uncertainty and decides variants or areas to sequence. The goal of this approach is to fill in gaps in coverage, for example. Weilguny tested the algorithm for dynamic enrichment to a mixed community sample. This led to a more homogenous coverage and more accurate variant calls. Over time, adaptive sampling using this approach rejected sequences of the most abundant species to allow for sequencing of the less abundant members of the community. The next step is to create a reference-free dynamic, adaptive sampling approach to eliminate the need for an input genome. They tested their model with the ZymoBIOMICS mock community and were able to enrich the least abundant members of the mixture and generate more complete metagenome assembled genomes (MAGs.). Wow! This new algorithm and approach will be powerful. I wonder if this approach can be used for the metagenomics course and compared to “traditional” MAG output.
