Skip to main content
Premium Trial:

Request an Annual Quote

New Methods Draw Single-Cell, Spatial Genomics Data Closer Together


NEW YORK – Single-cell and spatial genomics analysis methods are converging fast and may soon be indispensable for analyzing complex tissues.

That's one takeaway from last week's Biology of Genomes meeting, hosted virtually by Cold Spring Harbor Laboratory due to the COVID-19 pandemic. Several presentations used both spatial and single-cell genomics methods to provide insight into specific tissues of interest.

"Each method on its own has downsides that the other can compensate for," Andrew Adey, a researcher at Oregon Health Sciences University, told GenomeWeb following his presentation on sciMAP (single-cell combinatorial indexing from microbiopsies with assigned positions), a method that combines 100 micron "microbiopsies" of tissue sections with single-cell assays and combinatorial indexing to increase throughput. "Single-cell RNA sequencing is certainly common, and I see the same thing happening with spatial [genomics]. And both together will be the preferred path."

Another presentation showed the convergence coming from the spatial side. Lieber Institute for Brain Development researcher Andrew Jaffe presented on his lab's use of Visium, the new spatial transcriptomics platform from 10x Genomics, showing how the method could capture gene expression for single cells in some cases in certain brain tissues.

"We showed that a substantial fraction of spots contained only a single cell body, particularly in the less dense superficial layers of the cortex," Jaffe said in an email. "We did this by segmenting the images and counting the number of cell bodies that fell into each barcoded spot."

Both Adey and Jaffe's groups have posted BioRxiv preprints on their methods. Other talks covered similar themes but the presenters declined permission to report their work.

The presentations point to an imminent ability to capture genomic information about cells in heterogeneous tissues of interest. Other platforms, such as ReadCoor, are promising single-cell spatial resolution. And while Adey's presentation focused on a single-cell assay for transposase-accessible chromatin by sequencing (ATAC-seq), he said sciMAP could incorporate several other single-cell techniques, including RNA-seq, whole-genome sequencing, methylation sequencing, and Hi-C.

In his presentation, Adey was quick to note that researchers can pick from several existing technologies to provide spatial information, depending on what kind of biology they're interested in. "We kind of went down the rabbit hole on different techniques," he said. To investigate intracellular mechanics, which requires sub-micron resolution, researchers could use in situ methods, like seq-FISH or Merfish, or even electron and super-resolution microscopy for proteins.  Cell-cell interactions requiring resolution of several microns would be done with high-definition spatial transcriptomics, slide-seq, or any of several imaging methods. Many of those techniques can also be used to analyze lesions in tissues or other features relevant to pathology, but Adey noted there aren't a lot of options for analyzing epigenetics at this scale.

In consultation with pathologists, Adey said, his team decided to obtain microbiopsy punches between 100 microns and 500 microns across from frozen tissue sections. Each section biopsied is paired to an adjacent thinner one used for staining or other imaging to help correlate sites.

Biopsies were put on 96-well plates where nuceli were isolated using a publicly-available protocol and a transposase introduced the first indexing barcode. The nuceli were then pooled and redistributed to PCR plates, five nuceli to a well and barcoded again, giving each cell a high probability of having a unique barcode pairing. The combinatorial indexing process was developed in collaboration with Illumina in Jay Shendure's lab at the University of Washington, where Adey was once a graduate student.

Adey's group, led by graduate student Casey Thornton, validated the sciMAP preparation with single-cell ATAC-seq in human post-mortem brain tissue, looking at layers of the outer and inner cortex. They got spatial resolution of 200 microns with 200 micron spacing and with automation that spacing could get down to 5 microns. Each punch would have multiple cells within it. Additionally, sections can be stacked, providing spatial information in three dimensions. Based on the accessible chromatin profiles, excitatory neurons showed a gradient from outer to inner cortex, Adey said in his presentation.

Jaffe's presentation also focused on cortex tissue in the human brain and the lab's use of Visium to capture the transcriptome. Visium's "spots" used to capture transcripts are about 55 microns across, so not yet to single-cell resolution. Jaffe noted that his lab used single-nucleus RNA-seq datasets as a reference for gene expression from the same region. About 15 percent of all spots per sample contained a single cell body, he noted, and in some layers with lower cell densities, as many as 20 percent of spots contained one cell body.

About 10 percent of spots lacked any cell bodies. Some of these zero-cell spots still generated expression profiles, suggesting that they contained neuron "processes" or other appendages that cannot be easily analyzed with single-cell or -nucleus methods.
It should come as no surprise that as spatial methods mature, they're being rapidly adopted. Earlier this week, 10x Genomics CEO Serge Saxonov said the Visium platform has been installed by about 400 labs. Adey said combinatorial indexing methods have been adopted to some degree, hindered by the challenging workflow and the fact that the transposase used in the first round of indexing needs to be homebrewed. However, making reagents available as a kit is "in the works," he said, though he declined to comment further on any commercialization efforts for sciMAP or other sci-based assay methods. His lab has not pursued patents on their custom microbiopsy punch method, he noted.

Once it becomes easier to index samples, sciMAP could prove to be easily adoptable. "You can target whatever you want," Adey said. "At its core, it's a simple, straightforward platform and so versatile … you can plug and play a lot of options."