Towards real-time targeting, enrichment, or other sampling on nanopore sequencing devices, rather than in the sample prep: adaptive sampling for selective nanopore sequencing

Two new papers use real-time nanopore sequencing and the ‘Read Until’ API to show dynamic workflows for targeted sequencing without the need for upfront sample preparation. The papers describe methods that enable real-time electronic selection of molecules of interest, from a mixed sample.

Pre-prints from Kovaka et al. and Payne et al. published today utilising the ‘Read Until’ API for nanopore sequencing, showcase the potential future of rapidly sequencing regions of interest with nanopore technology, without the need for a target-specific sample preparation.

Adaptive sampling

Figure 1: Adaptive Sampling is a mechanism by which a user can programme their system to accept or reject strands based on a configuration specified in software.

Kovaka et al. developed a novel raw signal mapping algorithm that enables adaptive sampling to enrich sequencing of a collection of 148 human genes associated with hereditary cancers to 30X coverage on a single MinION flow cell. This enabled robust structural variant detection with more than double the number of SVs detected than with 50x coverage whole-genome short-read sequencing, as well as accurate small variant calling and DNA methylation profiling. Read pre-print.

Sam Kovaka noted “The ability to comprehensively detect genetic and epigenetic changes in any genomic region using a simple low-cost sequencing run will be transformative to cancer treatment and prevention, infectious disease profiling, and many other applications.”

Meanwhile, Payne et al. – the team that previously demonstrated this method in 2016 – have now shown multiple examples of scalable adaptive sampling, noting that direct base calling using Graphical Processing Units (GPU) is sufficiently fast to enable selective sequencing and that this scales to large references. The paper describes:

  • enrichment for a human exome panel covering over 10,000 genes.
  • a pipeline enabling enrichment of low abundant organisms within a mixed population without a priori knowledge of the composition of the sample.
  • sequencing of a targeted panel from the COSMIC gene panel including 717 genes. Including the identification of the known PML - RARA fusion event in the NB4 cell line in less than 15 hours of sequencing. Read pre-print

Alex Payne noted: “Read Until enables fast and flexible adaptive sequencing of many regions of large genomes, or specific subsets from multiple genomes. By reducing the time and cost of both sequencing and sample preparation these methods will allow researchers to focus long read sequencing on specific regions to address biological questions.”

Targeted sequencing, or selecting regions of interest from a sample, typically involves significant manipulation of DNA prior to sequencing. This can involve PCR steps, Cas-mediated enrichment or hybridisation capture, each of which can add hours or days to the preparation and hands-on time, as well as additional cost. By selecting molecules in real time during the sequencing process, these steps are negated, making the process simpler, maintaining DNA modifications present in the sample, and accelerating delivery of answers to biological questions.

Addendum:  10 February 2020

De Maio et al from the Goldman Group today released BOSS-RUNS, providing further insights into the possibilities of adaptive sampling with nanopore sequencing.  The group presents a new mathematical model and algorithm for the real-time assessment of the value of prospective fragments.  The decision framework is based not only on which genomic regions are a priori interesting, but also on which fragments have so far been sequenced, and so on the current information available regarding the genome being sequenced. As such, this strategy can adapt dynamically during each run, focusing sequencing efforts in areas of highest uncertainty (typically areas currently low coverage).

Nick Goldman noted:  "DNA sequencing can be a wasteful process, with time and resources within a sequencing run used to read and re-read regions that are of little interest or have already been sequenced to sufficient accuracy. For nanopore sequencing, our algorithm provides a practical way of focusing on just the DNA that you need to read, dynamically updating its strategy in the light of what you've seen so far. This is the first time that dynamic sampling using within-experiment feedback has been proposed for DNA sequencing.  The algorithm builds on genome location-based Read Until strategies of Kovaka et al. and Payne et al., combining them with a measure of the benefit of each possible successive read and an optimal strategy for choosing which fragments to read and which to reject and replace. This can considerably increase the efficiency of sequencing, improving the accuracy of genome reconstruction and reducing the time-to-answer."

Applications, and potential applications, of adaptive sampling

Adaptive sampling enables a large number of applications to be carried out with very basic sample preparation, leaving the complex target selection to be performed by the sequencer itself.

As demonstrated by these two papers, the 'Read Until' API can be used to enrich for strands that contain a target region of interest, thereby significantly decreasing the costs while preserving all the benefits of long-read sequencing. Users can also use it to reject strands from an organism which is of no interest. For example, in the case of microbiome applications, this could provide a simpler workflow, negating any need to deplete the host during sample preparation.

Another use for adaptive sampling is to balance coverage, barcodes or amplicons, ensuring target depths are achieved uniformly for the region(s) of interest.

Utilising the 'Read Until' API

The 'Read Until' API will be made available as a research tool under limited support in the Nanopore Community. The teams are now working on a new, simplified Adaptive Sampling API .For more information, visit the post on the Nanopore Community (login required)

If you’d like to be kept up to date about the Adaptive Sampling API from Oxford Nanopore, register your interest here.


Targeted nanopore sequencing by real-time mapping of raw electrical signal with UNCALLED

Nanopore adaptive sequencing for mixed samples, whole exome capture and targeted panels