Two research teams have independently published microbial metagenomic sequencing studies that used Life Technologies' Ion Torrent PGM, showing that the platform is a cost-effective alternative to longer-read sequencers for this application.
One team, led by Dag Harmsen at the University of Münster in Germany, sequenced amplicons from the V6 hypervariable region of the 16S rRNA gene in plaque samples taken from patients with chronic periodontitis. According to the authors, who published their results in PLoS One this month, this is one of the first studies using the PGM for amplicon-based metagenomics.
The other team, led by researchers at the University of Western Australia, analyzed samples from methanogenic waste treatment systems, both by sequencing amplified 16S rRNA and by shotgun sequencing, and concluded that "the PGM platform provides a low-cost, scalable, and high-throughput solution for both tag sequencing and metagenomic analyses." They published their results online in the Journal of Microbiological Methods last month.
The ultimate goal of the German researchers is to develop a diagnostic test to help predict the success of different types of treatments — in particular, antibiotics in addition to mechanical treatment — for chronic periodontitis, based on the presence and distribution of bacteria in gum pockets. Some of these bacteria are strongly associated with periodontal infections.
For their study, they analyzed the microbial community in subgingival plaque samples from four periodontitis patients before and after treatment. Two of the patients were conventionally treated by mechanical cleaning of gum pockets, the other two received conventional treatment plus antibiotics.
After PCR-amplifying the 80-base pair hypervariable V6 region of the 16S rRNA gene, the researchers sequenced DNA from each sample on one 314 chip with 200-base reads, generating on the order of 600,000 reads per sample, or between 5,000 and 8,500 high-quality unique reads.
They then performed an operational taxonomic unit, or OTU, analysis for each sample, which indicates the number of bacterial species, and found between 1,600 and 2,600 non-singleton OTUs per sample. After treatment, that number remained more or less unchanged, but the different types of bacteria were more equally distributed, and periodontal pathogens were no longer dominating.
"If this holds true with a larger sample size, then you can essentially use this for diagnostic monitoring of treatment success or failure," Harmsen, head of research in the periodontology department of the University Hospital Münster, told In Sequence.
The researchers also annotated high-quality reads taxonomically at the genus level and found that they could assign them to 330 genera, 38 of which were present in all samples.
According to the authors, the main advantages of the PGM for this kind of study are its small size, low acquisition prize, and fast turnaround time. Consumables costs are currently about $200 to $300 per sample, Harmsen said, which is about ten times less than for the Roche 454 platform, but "still a little bit too expensive for routine diagnostics." He said he expects costs to come down within the next year or so to a level "where it's possible to use this in patient care."
But the technology still has its limitations: For example, due to the limited read length, the scientists could not annotate the reads down to the species level but only to the genus level. Longer reads with fewer errors will allow them to sequence adjacent hypervariable regions and describe the microbial communities more accurately, they wrote.
"Of course if you have longer read lengths, it's easier to assign taxonomic levels because [the data] is more information-rich," Harmsen said. "So read length really matters." He said the 400- to 500-base reads from the Roche 454 platform are "definitely an advantage," but the price per experiment is ten times higher, and "if you consider doing many samples, price really matters."
He expects error rates for the PGM to go down with the 400-base pair read kits that Ion Torrent expects to launch later this year.
Harmsen said that the current study was proof of principle for a larger project he and his colleagues are planning. All samples came from a clinical trial with more than 500 periodontitis patients who were monitored over two years, with five samples taken from each. The scientists now want to analyze several hundred of these samples using the PGM, comparing patients undergoing different treatments or patients failing treatment.
Microbial Ecology: Piggery Waste
The Australian team used the PGM in a microbial ecology project, studying bacterial and archaeal community dynamics in a covered anaerobic pond, which is used to treat piggery waste and to generate methane.
For their analysis, they took monthly samples from the pond over a 10-month period. To study bacteria, they sequenced either 100-base amplicons from the V6 region or 200-base amplicons from the V3 region of the 16S rRNA gene. For archaea, they used a nested PCR approach to amplify 16S rRNA genes and sequenced those. In addition, for a few samples, they sequenced total genomic DNA. Sequence data were generated on the 314 chip or the 316 chip, using read lengths of 100 base pairs or 200 base pairs.
In total, the researchers generated about 350,000 filtered reads for each 314 chip and about 1.2 million filtered reads for each 316 chip. They assessed the read quality and found that the V6 fragment yielded about 80 base pairs with a quality score between Q20 and Q25, depending on the read length used. V3 fragment reads had a Q25 quality score with the longer read chemistry but a lower quality score with the 100-base chemistry. Overall, quality scores were "equaling or approaching those obtained for amplicon sequencing using 454 and Illumina-based technologies," the authors wrote.
They noted that in certain regions, about 40 base pairs from the amplification primer, the sequence quality declined and recovered over an interval of 10-15 bases. "We suggest that a systematic comparison of this and other platforms with the same sequence target may reveal the nature of these sequence quality variations as has been performed for other platforms," they said.
Using the QIIME pipeline for metagenomic data analysis, the researchers were able to classify reads to the bacterial root only using the shorter V6 fragments, and to known taxa — usually the genus level — using the longer V3 fragments. Thus, the V3 or the adjacent V4 region are "probably the best candidate[s] for streamlined analyses of microbial communities with the PGM through established pipelines such as QIIME," they wrote.
They also explored multiplexing 20 samples per 316 chip, recovering about 50,000 filtered reads that were suitable for subsequent phylogenetic analysis. "Balancing the need for more environmental sampling with the need for sequence coverage suggests that for multiplex analyses (with the possible exception of rare event analyses) it would be possible to multiplex up to 100 tagged PCR samples on a single 316 PGM chip at approximately USD$5 per sample at a coverage of approximately 10,000 reads per sample," they wrote.
Overall, the data were "of sufficient quality" for downstream analysis using pipelines such as QIIME, they concluded. The sequencing consumables cost is currently about $200 using the 314 chip and $500 using the 316 chip, they wrote. "Thus, the number and quality of reads are comparable with current 454/GS-FLX/Titanium technologies, but can be obtained for less than one tenth of the reagent costs."
"Whilst not currently on the same scale as newer generation ultra-high throughput technologies such as Illumina HiSeq and MiSeq, the future expansion, to include 10 million (318 PGM chip) and subsequent 500 million read chips, together with 400 bp chemistries, will make the PGM a valuable addition to the microbial ecologists' suite of tools," they said.
"We envisage this will primarily take the form of cost-effective and rapid-routine laboratory sequencing (e.g. 314 PGM chips), replacing relatively laborious analyses such as [denaturing gradient gel electrophoresis] and [terminal restriction fragment length polymorphism analysis], as well as providing small to medium laboratories with high throughput 16S rRNA and shotgun metagenomic capabilities (e.g. 316 and 318 PGM chips)."