typemat12.bsky.social
Microbial (meta)genomics researcher @sangerinstitute
interested in genomic epidemiology and evolutionary dynamics of STIs/NTDs/AMR in complex samples. Views my own.
https://www.sanger.ac.uk/person/beale-mathew/
73 posts
505 followers
490 following
Prolific Poster
Conversation Starter
comment in response to
post
And @carmcdhume.bsky.social recently did a blog about our field work using this method in Zimbawe sangerinstitute.blog/2025/02/17/s...
comment in response to
post
Although here we have used T. pallidum as a model, we think this approach has broad applications for larger pathogen genomes (bacterial, fungal, parasite), and we hope others can make use of these approaches for different pathogens.
comment in response to
post
We also built a @nextflow.io pipeline for rapid data processing, and a prototype shiny app for interactive data interpretation.
comment in response to
post
And we then deployed the method to a low-resource laboratory with our collaborators @thruzim.bsky.social BRTI in Zimbabwe as part of an ongoing genital ulcer study. The final assay can be performed in <2 days and costs <£13/sample.
comment in response to
post
We designed a 59-amplicon multiplex primer scheme which performs with high sensitivity (<qPCR Ct 32), with sequencing performed on MinION Flongle cells (24 samples per run), and high concordance between ONT amplicon SNP calls and Illumina WGS.
comment in response to
post
With novel (previously undescribed diversity), it functions more like a highly sensitive MLST - randomly occurring SNPs will occur by chance in amplicons, but we have many more amplicons than a typical MLST scheme so we have more opportunity to detect novelty.
comment in response to
post
This enables high resolution reconstruction of a WGS phylogeny using a small % of the genome (3.6% of T. pallidum genome here). The method recovers known sublineages (<20 SNPs apart) with high precision
comment in response to
post
We can use this to identify information rich regions, and then hierarchically select the optimal combination of regions to ensure coverage of all sublineages. We demonstrate this using a 59-amplicon scheme for Treponema pallidum (causative agent of syphilis)
comment in response to
post
When we identify SNPs discriminating individual finescale sublineages (e.g. <20 SNPs apart), the SNPs are scattered along a bacterial genome. However, if we consider discriminatory SNPs for many sublineages together, by chance we find positional clustering
comment in response to
post
Genomic pathogen surveillance is costly & challenging but multiplex amplicons eg @joshquick.bsky.social primalscheme works for viruses. Applying to larger (bacterial) genomes means compromise. For known pop structures, we identify key ancestral SNPs, then find regions which maximise discrimination
comment in response to
post
Think it's a week
comment in response to
post
Congratulations!
comment in response to
post
Looks great David, can't wait to read it.
comment in response to
post
Could authors point to a preprint as the preferred 'version of record', if they have to alter the version accepted at a journal?
comment in response to
post
Could try Horrible Histories? I think the history is generally very good, but it perhaps sheds light on things that the test authors don't want to cover?
comment in response to
post
Most UK born people don’t know most of the stuff in that test (including me). It’s often not part of popular culture (or school education) so not easy to absorb naturally. I think most people i know (inc my wife) just had to cram ahead of the test.
comment in response to
post
Basically nothing, but recipes built by the community (inc on bioconda) can include them anyway.
comment in response to
post
At least you still have conda access. The institutional licence changes were so high that our access has been blocked to default channels completely to stop us getting charged, which then effectively blocks most prebuilt package recipes.
comment in response to
post
Very cool, and nicely complements this www.biorxiv.org/content/10.1...