AGBT 2019, And the Winners are “Spatial Genomics, Single-cell Genomics, and Long-read Sequencing”

For the first time in three years AGBT returned to Marco Island for its 2019 conference. It was pleasant to be back and be able to enjoy the warmth and the friendly setting we’ve all remembered. The weather played along nicely, after initial thunderstorm announcements, the beach was pleasant and the talks were – as usual – of outstanding quality.

The following paragraphs provide some talk and commercial product highlights as collected during the conference.

Spatial Genomics

Clearly, spatial genomics seemed to be the winner of the conference, as reflected in its representation as a critical topic across the many talks such as the ones from Sanja Vickovic (Broad Institute), Katherine McNamara (Stanford University), and many other presentations, as well as in new product announcements by Nanostring with the GeoMx Digital Spatial Profiler, and 10x Genomics – who will commercialize its spatial expression technology, the Visium product,  in the second half of 2019, based on the recent acquisition of Spatial Transcriptomics in December of 2018.

The general methodology overlays sequencing data (RNA-seq expression data) across a tissue section (e.g. pathology tissue section) and therefore provides a deep look at the expression profile in the context of a morphologically conserved physiologically highly relevant tissue section which helps link observations to the specific function and cellular context. In addition to obtaining histopathology images, if properly scaled billions of new data points can be collected within a single experiment. Clearly, spatial genomics will aid in the understanding of subtyping specific cells based on their location within a tissue, with promising practical application such as understanding tumor heterogeneity.

• Sanja Vickovic (Broad Institute) – Genome-wide spatial mapping of gut host (mouse)-microbiome relationships – focused in her talk on spatial host (mouse) and microbiome transcriptome sequencing to study host-microbiome interactions in the gut. Taking advantage of a genomic-wide spatial analysis platform in combination with overlayed scRNA-seq data resulted in 3 billion data points (1,900 sections of spinal cord and 190K spots). It took Vickovic 1.5 years to develop a proper methodology to permeabilize bacteria while keeping spatial information – 16s and spatial distribution – accurate and to sequence the host RNA along with the bacteria. A cross validation she presented was in the area of ALS (spinal cord ventral horn) dynamics across time. She also touched upon various new applications such as Splotch (combines ST data and scRNA-seq data) and HSTD (High Density Spatial Transcriptomics). Vickovic highlighted that a key driver for adoption of spatial transcriptomics will be ease-of use.
• Nanostring communicated shipment of their GeoMx Digital Spatial Profiler (DSP) this June. GeoMx works with slides stained with oligo-conjugated antibodies to tag proteins and RNA molecules of interest. After selecting the region of interest, oligos are UV-cleaved off antibodies and aspirated for sequence and count analysis.

Single cell genomics

Recent advancements in single-cell genomics were reviewed in Aviv Regev’s (Broad Institute) talk who applied single-cell genomics to study ulcerative colitis and resistance to immunotherapy. Regev emphasized that there is a need of new tools for advanced applications of single-cell genomics. She followed up by highlighting the emerging landscape for spatial genomics, including Nanostring’s GeoMX DSP. Regev views spatial genomics as an extension of single-cell genomics and demonstrated the power of spatial genomics information to research cold versus hot niches in melanomas based on CD8 T-cell clusters. The result is the extraction of a “resistance signature” based on 31 patients in total.

Long-read sequencing

100 tomatoes in a 100 days

The potential of long read sequencing was well demonstrated in Michael Schatz’s (Johns Hopkins University) presentation who provided a detailed look at the structural variant landscape of tomato genomes, which were until recently hard to find using regular sequencing technologies. Schatz is now sequencing about 12 to 16 samples per week which translates into about 100 tomato genomes (there are ~15,000 varieties) in 100 days. This is made possible via the PromethION platform – he started out with MinION and GridION sequencing – which routinely achieves 80-90 GB (average of ~75GB) with single read accuracy of over 90%. Schatz emphasized that the field is progressing thanks to improved technologies and bioinformatics which results in increased throughput (comparable to the NovaSeq throughput) and lower cost! Here a link to the Schatz’ full presentation.

Other long-read sequencing related talks

And then of course there were the many applications in relation to long-read sequencing which included, but are not limited to talks by:

• Alex Kentsis (Sloan Kettering) who focused on understanding why young kids get cancer by applying long-read sequencing (using PacBio Sequel) which he stated produces better maps and insights into biology;
• Mike Hunkapillar (CEO at PacBio) who talked about the fidelity of long-reads for comprehensive genomic analysis (instead of an awaited discussion of how PacBio plans to merge with Illumina); or
• Karen Miga (UCSC) who showed how to get eleven reference quality genomes in only nine days on three Nanopore PromethION flowcells using sampling to maximize allelic diversity and the Circulomics short read eliminator kit to obtain 100kb length reads. Miga emphasized to parallelize long-read sequencing with consistent scalable assembly requires the right computing and phasing capabilities. The PromethION workflow includes phasing, which Miga considers a critical component. With this set-up, across all flow cells the average is 50-100kb length reads which is meeting the team’s needs for sequencing coverage. As of now they are completing 10 reference genomes in 10 days, but they still need to reduce the cost to reach their goal of $10K for 10 genomes.

Parasitic nematodes talk by Elodie Ghedin (NYU)

Filarial nematodes were front and center in the lively Ghedin talk. Her motivation to research these worms/parasites is because of their immunomodulating capacity as they secret metabolites and glycoproteins to downregulate the host immune system which allows them to hide for years from the immune system in plain sight. She highlighted the sequencing nightmare she had to overcome to sequence two tiny organisms that are completely co-dependent and cannot be independently cultured, and how the PacBio technology came to the rescue.

The Wolbachia intracellular bacteria, an endosymbiont, within the filarial nematode Onchocerca volvulus – the causative agent of “river blindness” – represents an incredibly complex system to culture and is challenging to genomically dissect. Ghedin pointed out that it is difficult to use non-parasitic nematodes, like C. elegans as a model, as they do not translate well to other systems for understanding what is going on at the genome level. Besides using PacBio for whole genome assembly, Ghedin also introduced the exciting new frontier of spatial methods, including Insi2vec and the GeoMx DSP to further dissect these host nematode systems.

Coinciding product launches/announcements worth mentioning

• Genapsys discussed Early-Access Customer Data for their sequencer and the launch of the new platform later this year. The new platform is an iPad-sized, short read electronic sequencer with Illumina-like accuracy for about $10k.
• Agilent announced a new library preparation system for next generation sequencing. The new Magnis NGS Prep System is designed to run complex DNA sequencing assays fully automated. It includes reagents and protocols that make it easy to assay multiple genes and complex genetic aberrations from genomic DNA, including degraded samples such as formalin-fixed paraffin embedded (FFPE). 
• Lexogen launched CORALL, an all-in-one total RNA-seq library prep kit for all whole transcriptome analysis applications such as differential expression, isoform quantification, single nucleotide polymorphism (SNP), and mutation detection.
• 10x Genomics announced that it will commercialize a spatial gene expression technology in the second half of the year. 10x’s Visium Spatial Gene Expression Solution will be based on the technology originally commercialized by Science for Life Laboratory spinout Spatial Transcriptomics, which 10x acquired last year.
• NanoString presented customer data from its Digital Spatial Profiler platform, GeoMx DSP, which it plans to launch at this month’s the American Association for Cancer Research (AACR) meeting. The will ship machines in June of this year.
• MGI Americas (formerly BGI) ended the meeting with providing details on their new professional-refrigerator sized sample-prep-to-VCF sequencing machine, announced October 2018 and called MGISEQ-T7, which is powered by the DNBseqTM technology.