Skip to main content
Premium Trial:

Request an Annual Quote

Two Research Groups Develop Droplet Microfluidic Techniques for Single-Cell Sequencing


SAN FRANCISCO (GenomeWeb) – Two research groups have developed techniques that harness microfluidics for single-cell sequencing.

In a study published in Nature Biotechnology this week, researchers from the University of California, San Francisco describe a droplet microfluidic method to sequence genomes of single cells from environmental samples. The team, led by Adam Abate, has licensed the method to the startup Mission Bio. Separately, a group from the New York Genome Center has developed a droplet microfluidic device for single-cell RNA sequencing, which they described in a paper on the preprint server BioRxiv. 

The publications underscore a growing interest in single-cell sequencing technology, particularly high-throughput, low-cost methods.

"My belief is that in the near future, basically all sequencing of living organisms will be done at single-cell resolution," Abate, principal investigator in the department of bioengineering and therapeutic sciences at UCSF, said.

Abate's group and the NYGC team both applied droplet microfluidic technology to address different challenges related to single-cell sequencing. Abate's team wanted to focus on a method that would be scalable to tens of thousands of cells and would sequence DNA, rather than RNA. Meantime, the NYGC team looked to address the cost of instrumentation, and demonstrated their device in their paper using the Drop-seq method, a single-cell RNA sequencing method originally developed in the Harvard Medical School laboratory of Steven McCarroll.

Rahul Satija, a core member and assistant investigator at the NYGC, and a senior author on the BioRxiv paper, declined to comment on the method since the group has submitted the publication to a peer-reviewed journal. In the manuscript, the researchers reported that they designed and assembled a microfluidic device from 3D-printed parts for $540. They demonstrated that they could use the device in combination with Drop-seq to sequence transcriptomes from more than 8,000 single cells from the synovial fluid of a rheumatoid arthritis patient.

In the Nature Biotechnology study, the UCSF team made use of hydrogel microspheres, or microgels, that trap DNA molecules but are permeable to enzymes and small molecules. That enabled the researchers to first capture single microbes in a microgel and then run a series of washes to purify and fragment the genomic DNA.

The cells are captured in the microgel using a "two-stream co-flow droplet maker," the authors wrote in the study. Essentially, a cell suspension stream is merged with an agarose stream to form a droplet.

After purifying and fragmenting the DNA, the researchers used a microfluidic device to merge the microgels with a droplet containing PCR reagents and a droplet containing a barcode.

The contents can then be pooled and sequenced. The setup is able to process more than 50,000 single cells in a few hours.

The researchers first validated that their approach was able to capture and sequence genomes from single cells by applying it to an artificial microbial community of 10 species, including gram positive bacteria, gram negative bacteria, and yeasts.

Sequencing on the Illumina MiSeq resulted in 6 million reads that could be divided into more than 48,000 barcode groups, representing between .1x and 1x genome coverage. Abate acknowledged that the low genome coverage was a drawback of the method, but said that the team is now working on improving this.

Next, the researchers mapped the reads to the reference genomes of the 10 species to determine whether the barcode groups corresponded with single cells, assigning purity scores to each group. For instance, if two microbes ended up in the same barcode group, reads would map to two different references, resulting in a lower score. The ideal purity score would be 1 and the researchers found that the majority of scores were greater than .95. Further inspecting those with lower scores, the scientists found that the most common cause was two cells being captured in the same droplet or two droplets merging.

The team then wanted to confirm that the method gave an accurate representation of the abundance of each species, so they compared it with other methods, like microscopy, and found that it gave a comparable estimate. And finally, the researchers evaluated coverage biases caused from uneven amplification, finding minimal bias, which they attributed to the small reaction volume, which helps reduce biases.

The researchers also developed a method for analysis, which they called in silico cytometry. To do that, they first built a database of the sequenced reads, which they called SiC-Reads, organizing the reads as barcode groups. Then they could use that database to look for correlations. Abate said that this analysis enabled them to make the most of the low-coverage genomes. "Even though the genomes are low coverage, you still have information about which sequences are linked together in a single genome, which is fundamentally different from metagenomic data," he said.

After validating the method, they tested it on an environmental sample collected from coastal seawater, generating around 8 million reads. Phylogenetic profiling classified around 7 percent of the reads into bacteria, archaea, and viruses. Next, they built the SiC-Reads database and used the in silico cytometry method to analyze the data further. For instance, Abate said one goal was to look for genes that confer antibiotic resistance. They were able to search for known antibiotic resistance genes and then use the barcodes to figure out which species had those genes. In addition, the team could do a similar search for virulence factors and figure out whether species had both resistance genes and virulence factors, "a combination that can be bad for humans," Abate said.

The team is now working to improve the method. The primary drawback, Abate said, is that it results in low genome coverage. One way he is looking to improve that is to pre-amplify the DNA using isothermal amplification.

In addition, startup Mission Bio, based in San Francisco, has licensed the technology and will commercialize a similar version to what was described in the Nature Biotechnology study, according to the company's CEO Charlie Silver, who declined to disclose further details.

Abate said that aspects of the molecular biology would have to be slightly different for Mission Bio's purposes, which would focus on human cells rather than bacteria. "But fundamentally, the technology is the same," he said.

Meantime, in his own lab, Abate will continue to focus on applying the method to bacteria. "The scale of diversity of bacteria is staggering and has been a major challenge in understanding bacterial ecologies," he said. The sheer number of cells present in a small sample "is the main challenge to understanding the ecology," he said, and better methods are needed that can connect genomic sequence data to individual bacterial cells.

Abate added that although there are now a number of commercial single-cell sequencing technologies on the market, they all have drawbacks, which he was looking to address with the SiC-seq method.

For instance, he said, 10x Genomics' technology, which also uses droplet microfluidics, currently only works for single-cell RNA sequencing, rather than DNA sequencing. "RNA is already naturally fragmented, and so is easier to capture," he said.

Other researchers have been looking to harness the Hi-C method for single-cell sequencing, including researchers from the University of Washington and the Babraham Institute in the UK, as well as startup Phase Genomics.

Abate said that such an approach would be a competing method.Although the Hi-C method "is a lot simpler and more accessible for non-experts to adopt," he said, there are also limits to its efficiency and genome coverage. Plus, while the droplet microfluidic technique may be more complicated to develop, it can more easily be made into a commercial product, he added.