Sequence statistics to report viruses (or bacteria and fungi) using Nanopore for diagnostic purpose

Question

[pixie@bioinfo]

Hello,

Currently, our diagnostic lab uses nanopore sequencing for virus detection. Sometimes, we are in the grey zone while assigning a specific virus species. Also, not all viruses are equally represented in the NCBI databases.

Other that the number of reads mapping to a viral genome, visualization on Jbrowse for the contigs and subsequent sequence similarity NCBI, what other metrices can I use to help them report these cases ?

Currently we are using Pavian for reporting ( https://ccb.jhu.edu/software/pavian/)

Thanks

Answer 1

[colindaven]

Viruses are quite hard to accurately quantify in metagenomic sampling, since they are so rare relative to bacteria and the host (you probably use enrichment, but didn't say which type). It might be useful to have multiple scans of representative viruses (ie. those that are not identical or a couple of base pairs distrinct from one another) within each clade, since if you use all viruses in NCBI then you can't use mapping quality filters, since they will mask each other.

Maybe you need to cluster the reference sequences and use only one representative per cluster at 95% identity etc.

I liked this approach when I looked at it some time ago, it might be helpful for you.

https://github.com/esteinig/vircov

Generally, I would recommend using multiple metagenomic programs for each positive diagnostic hit.