28th June 2025
New pre-print out in VeriXiv, link: https://verixiv.org/articles/2-146/v1
We produced data analysis tools for identifying haplotypes (combinations of genetic variants that occur together) from nanopore sequencing reads.
Our machine learning models can be trained to recognise a set of important haplotypes at genetic loci of interest, such as drug susceptible vs. resistant sequences in a drug resistance marker gene. The model then compares each individual sequence read against its library of expected haplotypes to assign which haplotype (if any) each read belongs to. That way, we can assess the frequency of different haplotypes in a sample - this is especially useful for mixed clonal samples where multiple different clones possessing different haplotypes are present in the same sample - a common feature of malaria parasite infections termed multiplicity of infection. Mixed clonal samples are seen in other pathogens too, eg drug resistant and susceptible strains of bacteria like Pseudomonas can often co-exist in the same sample.
The model can be trained to recognise any sequence from any organism - in the paper we illustrate its use with malaria parasites and SARS-CoV-2, the virus that causes Covid19.
​
The model is freely available to try out from GitHub:
https://github.com/paopaoch/VariantCalling
​
We designed an accompanying pipeline to import nanopore sequence data into Python for easier machine learning training, which we call PicklSeq - also available for free on GitHub:
https://github.com/paopaoch/PicklSeq
​
This was a really fun collaboration with some super talented students from the Cambridge University Institute for Manufacturing in the Department of Engineering.
​
Please give the code a try!
Variable rates of SARS-CoV-2 evolution in chronic infections - new publication!
28th April 2025
New paper out in PLOS Pathogens
This paper explores within-host population structure and evolution in chronic infections with SARS-CoV-2.
Covid-19 is generally an acute illness lasting a few days or short weeks; however, in some individuals the virus can persist for weeks or months. This can happen in people with weakened immune systems, leading to within-host evolution. We analysed nine cases of chronic SARS-CoV-2 infections and identified separately evolving viral sub-populations within infected hosts. These distinct sub-populations varied in their rates of evolutionary change, with some viruses evolving faster than others. The paper shows how complex within-host viral dynamics can be over the course of chronic infections, which may contribute to the emergence of novel viral variants, such as the Omicron variant that can then sweep across the globe. ​
'DRAG2' pre-print out on VeriXiv
13th March 2025
Pre-print out here: https://verixiv.org/articles/2-30.
Working with colleagues in the UK, Ghana and Cameroon, we have updated our nanopore-based amplicon assay for malaria parasite genetic surveillance. I called the original assay 'DRAG', for drug resistance + antigen. So, this updated assay has the super original name of DRAG2. It includes additional drug resistance markers compared to DRAG1, plus msp2 and a fragment of the 18S rRNA gene for species detection. We've included a lot of information on how the assay can be set up and run in a laboratory, including the use of plasmid inserts for use as positive controls. Hopefully this will be a useful assay for malaria genetic surveillance using nanopore sequencing! Funded by the Gates Foundation.
​