Oxford Nanopore Sequencers are rapidly gaining popularity because of their portability, ease of access, and ease of use. However their single read error rates are still high making applications such as multi sample processing, spatial and single cell RNA barcoding challenging. Flow cells that are used in these sequencers are also expensive, which prohibits them from being used more widely. Molecular barcoding helps sequence a variety of samples concurrently but suffers from high single read error rates. We previously developed a tool to demultiplex 4 molecular barcodes named DeePlexiCon that addressed some of these problems and is able to demultiplex barcoded samples with a very high recovery and accuracy [1].
In this talk I will introduce a new tool called "Total Recall" that not only allows instant recognition of barcodes directly from raw Nanopore signal, but also able to recognise whole or partial gene sequences using the signal. The system potentially can be trained to a variety of libraries to cater for subsample recognition or can be trained to utilise a function of Nanopore sequencers called "read until" to selectively accept or reject samples potentially offering programmable gene panels. Combined with multi-sample barcoding this tool can help the promise of Nanopore sequencers closer to reality by enabling them to process multiple samples selectively and giving near instant results on the sequenced data.