Update from Day 2 of the Nanopore Community Meeting in New York: Cas9 enrichment, long reads in transcriptomics, fungal pathogen outbreak surveillance, more
The 200 attendees returned after a busy Day 1 to a day of plenaries, breakouts and discussions.
Dr. Andy Heron, Oxford Nanopore's Director of Advanced Research, told us about the progress they've made in target enrichment using the CRISPR/Cas9 system on the nanopore platform and some of the unique ways this can implemented on a nanopore system to create relatively simple and fast sample and library preparation workflows for in-field sequencing and detection applications. Target enrichment enables many different workflows, and to work best with nanopore sequencing they want it to be both simple and fast. Andy covered work to improve sensitivity in an effort to reduce the input requirements of the system, an aspect that is crucial to detecting the minute quantities of DNA that result from specific enrichment. He explained how enrichment is all about getting DNA to the pore, described how beads are an ideal delivery mechanism and how they can increase the throughput by an order of magnitude. He then went on to describe how they're using Cas9 to enrich specific targets for sequencing.
Cas9 can bind very quickly at room temperature with high specificity, by adding a tether to Cas9 they could then deliver the target DNA to the pore. Using this method they achieved highly sensitive and specific enrichment of several E coli genes even against a background of human DNA. You can also multiplex the samples which increases throughput
Andrew Smith from University of California Santa Cruz presented recent work sequencing mRNA, 16S rRNA and tRNA on the MinION using the new Direct RNA sequencing kit currently being released to few developers. He described some unexpected results that suggest structural 16S rRNA may be damaged during reverse transcription. Andrew showed that long reads of 16S RNA sequenced directly allowed their group to make accurate bacterial identification of e.coli MRE600, V.cholerae A1552 and M. maripoludus str. S2 and potentially provide more information than sequencing a smaller subset of the 16S variable regions. He described some modifications to the ONT protocol that allowed sequencing of 16S rRNA directly from cell pellet in 60 minutes.
Breakout 3:
Sophie Zaaijer from the New York Genome Center presented a strategy for real-time person re-identification using the MinION. She presented a methodology her group developed that uses SNP profiling with sparse data to identify a person with 100% probability using a real database containing ~30,000 individuals and using only ~99 MinION generated SNPs. Re-identification can be completed in ~3 minutes. Sophie showed that they can use this methodology to identify a first-degree relative or identify contamination of other cell types in a prepared DNA sample. The whole protocol can be made portable using the rapid-kit library preparation in combination with running our re-identification pipeline. The method can be applied in remote areas, in the lab, or in the clinic and can be safeguarded by the database it is compared to.
Lachlan Coin from the University of Queensland presented a novel targeted sequencing method which allows simultaneous analysis of hundreds of repeats up to 2kb. He described using sequence capture along with VNTRTyper, a novel algorithm their group developed that utilises an HMM to identifying repeat copy number from long-read sequence data. They used PCR fragment length analysis to validate their results. This novel genotyping approach is a new cost-effective tool to explore a previously unrecognized class of repeat variation in genome-wide association studies of complex diseases.
John Tyson from University of British Columbia presented his work using long reads generated on the MinION to enhance the understanding within the Genomic/Transcriptomic space. He presented work looking at genome organization and single-molecule transcript analysis of long cDNAs in C. elegans. For example, he has identified the full splicing profile of transcripts by amplifying low-abundance splice variants.
Isac Lee from Johns Hopkins University presented his work that employs solution-phase hybridization-capture to target the p16 and SMAD4 gene regions, which have been shown to frequently contain SVs and SNVs in pancreatic cancer. He explained how his group leveraged the resulting high sequencing depth and computational correction of the reads (nanopolish) to perform both SV and SNV detections. The long sequence reads allow phased SNV detection of >2kb, which can be used to further study the effect of SNVs in cancer, such as allelic preference of cancer-causing mutations or the pattern of mutations.
BREAKOUT 4: Data analysis
Matt Loose from the University of Nottingham kicked off Breakout 4 with an update to his popular MinoTour software. MinoTour is available on github as a local installation and handles basecalled data from Metrichor, Nanonet and MinKnow. With throughput ramping up with the new chemistries, Matt has been working on handling the vast amounts of data produced by the MinION. New features include capturing of live data from MinKnow as well as monitoring features e.g. automated messaging when a run has reached a particular coverage or warnings when local storage may be running out while running the device. Matt also described his work with read-until, the ability to reject DNA molecules during sequencing that does not match a given target sequence. His work with Zam Iqbal on enriching TB data against the background host genome yielded impressive results.
Prof. Jeroen de Ridder demonstrated how his lab is using long reads to map the 3D structure of the genome. While we typically think of the genome as a string of letters, it has a 3D conformation in the cell and this conformation is important. For instance, it has been shown that changes in chromatin loops can lead to oncogene activation. One method to map the structure is called 4C, however this is based on short reads so can only map pairwise interactions. In order to get to higher order interactions you need long reads, they developed a method based on 4C that can capture up to 6-way interactions. Using nanopore data they've uncovered not only the existence of 'hubs' of DNA contacts but also potential competition between interacting sites.
Next up in Breakout 4, Fernando Izquierdo-Carrasco gave an overview of the EPI2ME platform, lifting the curtain and outlining the infrastructure behind the platform as well as what to expect application-wise in the near future. Fernando outlined the advantages of cloud-based bioinformatics, while highlighting EPI2ME’s low barrier to entry - as simple as uploading your files to the service - goodbye to dependency hell! EPI2ME provides users with out-of-the-box analysis applications complete with real-time updating eye-catching and detailed overviews of the processed data. Fernando spoke about the new 16S application as well as real-world examples of antibiotic resistance detection workflows in partnership with Tariq Sadiq and the importance of run-until when generating clinically actionable results.
Dr. Christiaan Henkel showed how they generated a draft assembly of the eel genome with 14X coverage of nanopore data. Christiaan is interested in the genomes of two organisms in particular: i) the eel because it occupies an interesting position in the tree of life and it's tasty and ii) the tulip genome because economically important and he's Dutch. The tulip genome will be particular challenging as it's 35Gb and contains a lot of transposons. Assembling it using existing methods would involve a prohibitive all-against-all read comparison, so they've developed an assembler aptly called Tulip. It replaces the all-against-all alignment step with an alignment against 'seed' sequences which can be generated from an existing genome assembly, a genetic map or nanopore reads themselves. Using this approach they were able to improve upon the existing eel genome assembly. Christiaan is planning to use PromethION to sequence the tulip genome.
On to cancer research, in the afternoon, Chia-Lin Wei from the Jackon Lab discussed their work using the MinION to detect whole-genome SVs in a TNBC cell line model. They established a sample-to-data workflow and customised bioinformatics pipeline for fast and accurate SV analysis, detecting common types of SVs with high specificity from minimal genome coverage. Thousands of genome rearrangements were uncovered affecting genes of relevant functions in their study and they also deciphered complex SV patterns overlooked by short-read analysis, revealing genome-scale breakpoints at base resolution. Dr Wei noted that “Our data supports the feasibility of using long-read sequencing for SV detection in cancer cohorts at real-time to trace clonal amplification, monitor drug sensitivity and outcome of therapeutic intervention.”
Johanna Rhodes Faculty of Medicine, School of Public Health, Imperial College London, UK Nanopore sequencing of a pathogenic fungi outbreak in a UK hospital. Last plenary speaker of the conference was Johanna Rhodes, early career fellow at Imperial College London, who showed how the MinION was used for effective fungal outbreak investigation and management in a UK hospital. Fungal infections kill over 1 million people a year and anti-fungal resistance is a growing problem with the newest discovered class of anti-fungal drugs seeing resistance within a year of introduction. Those with compromised immunity are particularly at risk. In April 2015 the first case of Candida auris was identified in a UK hospital. This case was part of a globally emerging multi drug resistant fungal outbreak. With a 60% mortality rate in those contracting it in hospital, it was crucial that this outbreak was managed and monitored closely. Johanna outlined the use of sequencing to generate a gold-standard reference and perform speciation and resistance profiling with the MinION. Johanna’s work shows how real-time long read technology is crucial for effective investigation and management of outbreaks, contrasting with slow traditional typing methods employed within the health-care system today. Highlighting the speed, portability and accessibility of the MinION, Johanna suggests that this technology could usher in the era of bedside diagnosis and personalised treatment.