We are very pleased to release DNAscent v4.0.1 today which is the first version of DNAscent designed to work on sequencing data from Oxford Nanopore's R10.4.1 flow cells. The usage and functionality is nearly identical to that of v3.1.2 which was the last release designed for Oxford Nanopore's legacy R9.4.1 flow cells. DNAscent v4.0.1 calls the location of the thymidine analogues BrdU and EdU across single molecules of DNA sequenced on R10.4.1, detects the speed and stress of DNA replication forks, and calls analogue-positive regions of each read.
As developers, R10.4.1 has given us a lot of new tools to play with (particularly the higher 5 kHz sampling rate). We've managed to convert these new features from the Oxford Nanopore platform into BrdU and EdU calling that is much cleaner than what we observed with v3.1.2 on R9.4.1. While we're very proud of v3.1.2, it could struggle a bit with confusing BrdU for EdU (and vice versa) which was something we were looking to improve. From our benchmarking so far, v4.0.1 has been able to capitalise on R10.4.1 to make a drastic improvement in this area where we observe much cleaner analogue calls. On human DNA that hasn't been treated with BrdU or EdU, we observe a false positive rate of approximately 0.05% on v4.0.1 (or about 1 in every 2000 thymidine bases) which is about an order of magnitude lower than on v3.1.2. This has given our fork-calling algorithm a much easier job to do.
We've been very pleased by how v4.0.1 has performed in-house, but we're also aware that a community has built up around DNAscent with users working on all sorts of applications outside of what we're using it for. This has been fantastic to see. As the R10.4.1 pore is still relatively new (particularly with the faster 5 kHz sampling rate) we've decided to designate v4.0.1 as a pre-release. We hope this will help the community to continue on with their science while allowing us to listen to community feedback as we improve things towards a long-term support release.
This has been a challenging development cycle and it has involved a lot of people coming together to build something that we think is really cool. On the engineering side, Pauline and Shutong in my lab both contributed code, testing, and a lot of help with just about everything. We're very grateful to Mathew Jones at the University of Queensland for providing the training data used in this release. We also want to thank our collaborators at Catherine Merrick's lab and Sarah McClelland's lab for test data and lots of really useful input as well as the MRC, BBSRC, and Leverhulme Trust for funding. Lastly, a big thanks to everyone involved for agreeing to release the software as soon as it was tested and stable so we can get it into your hands as soon as possible.
You can access DNAscent v4.0.1 here.
Lots of new things coming soon so watch this space.
Cheers,
Mike