Skip to main content
Premium Trial:

Request an Annual Quote

Public Health England Team Describes Metagenomic Sequencing Strategy to ID Viral Pathogens


SAN FRANCISCO (GenomeWeb) – Metagenomic sequencing is increasingly considered a valuable tool for infectious disease research, diagnostics, and outbreak surveillance. One significant hurdle, however, is the fact that a pathogen usually makes up a very small portion of total genomic material in a high background of human DNA and other microbial genomic material, so direct sequencing from clinical samples has been challenging.

Now, researchers from Public Health England have described a protocol for metagnomic sequencing of RNA viruses directly from clinical samples without amplification or enrichment, and have tested it on both the Oxford Nanopore MinIon and the Illumina MiSeq sequencers.

Steven Pullan, project leader of the genomics of rare and emerging human pathogens team at Public Health England, said that the protocol, described in a paper published on the BioRxiv preprint server in April, is similar to a metagenomic sequencing strategy for the MinIon originally described by Charles Chiu's group at the University of California, San Francisco. In Chiu's 2015 Genome Biology paper, the lab showed in a proof of principle study that a metagenomic sequencing protocol on the MinIon could identify a pathogen. But, Pullan said, that study was done in just a few samples, all of which had a high viral load, so the PHE team wanted to see whether it would also work in a larger collection of clinical samples with varied viral loads. 

Pullan said that his group was interested in using metagenomics and testing directly from clinical samples because its is focused on designing protocols that can work in the field in the midst of an outbreak. In such situations, culturing a sample before testing is not ideal because it is time consuming and requires more extensive laboratory equipment.

The group focused on the RNA viruses dengue and chikungunya, in part because it had access to patient samples for those diseases at the UK's Rare and Imported Pathogens Lab. "The most predominant samples there are dengue and chikungunya," Pullan said.

In the study, the researchers evaluated 26 samples from the lab that had all tested positive for either dengue or chikungunya. One key step in the protocol, Pullan said, is to degrade the DNA after extracting nucleic acids, which removes a lot of background material, making it easier to identify the RNA virus. Next, the team prepared cDNA using a sequence-independent primer amplification strategy, similar to that described by Chiu's team in 2015, and then performed a random reverse transcription step, followed by sequencing.

"The real surprise to us was the high proportion of reads that were viral," Pullan said, adding that the main reason for that was likely the DNA degradation step. Metagenomic assays are typically thought to produce a "tiny fraction of reads on target, so it was a shock that we could sequence samples and for some have the majority of reads be viral," he said.

Another UK team is working on developing methods to sequence patient samples directly in order to diagnose tuberculosis but has found that doing so is difficult due to the high background of human DNA. But since chikungunya and dengue are both RNA viruses, that background can be eliminated without losing the pathogen genome.

In general, although sequencing on Illumina resulted in a greater percentage of mapped reads from viral genomes, sequencing to 20x coverage on the MinIon resulted in similar genome coverage.

For instance, even in the chikungunya sample that resulted in the lowest percentage of viral reads, at just 5 percent, with MinIon sequencing, and 22 percent with MiSeq sequencing, sequencing to 20X coverage on the MinIon still resulted in more than 89 percent genome coverage.

Overall, Pullan said, both the MinIon and MiSeq were able to identify the virus and cover the majority of the viral genome. In addition, the researchers were able to identify that one of the chikungunya samples had a coinfection with dengue. Even though only .08 percent of MiSeq reads and .15 percent of MinIon reads mapped to the dengue virus, sequencing to 20X coverage on the MinIon enabled both the primary chikungunya virus and the dengue virus genomes to be covered at more than 99 percent and 95 percent, respectively.

Throughout the course of the study, Oxford Nanopore released two new library prep kits, the 1D2 kit and the rapid kit, which the PHE researchers tested as well. They noted that although both kits resulted in a lower proportion of viral reads, the 1D2  kit made up for that with an increase in total data to 5 million reads from 1.8 million reads with the 2D kit. Meantime, the Rapid kit had the benefit of a simplified and faster sample prep of just 10 minutes, the authors noted.

Going forward, Pullan said, the team plans to continue testing the various metagenomic protocols to figure out which technologies make the most sense for which applications. He added that the main focus currently is to use the protocol in the field for real-time outbreak surveillance, but that the lab would also consider developing a diagnostic test for dengue or chikungunya.

"There are different considerations for a diagnostic test," he said, that would have to be taken into account and tested. For instance, a diagnostic assay would likely include more control samples spiked in to assess background sequences and contamination. Also, studies would have to be done to determine thresholds for limit of detection, so it would be known how large a viral load would be needed to make a diagnosis. Furthermore, the level of multiplexing that could be done to balance cost while minimizing risks for cross-contamination would need to be determined.

In addition, he said, the group is involved in projects to develop metagenomic diagnostic assays for influenza from respiratory samples.