NEW YORK (GenomeWeb) – Before selecting a single-cell, whole-genome amplification strategy for a given study, researchers should carefully consider the genome features most crucial to the investigation at hand, according to a new quantitative analysis in PLOS One.
Stanford University's Stephen Quake and colleagues performed dozens of experiments on DNA from bulk or single-cell Escherichia coli samples amplified using either multiple displacement amplification (MDA), multiple annealing and looping based amplification cycles (MALBAC), or New England Biolabs' PicoPLEX single-cell WGA kit (manufactured by Rubicon Genomics).
After sequencing the resulting amplicons with Illumina's MiSeq instrument, the team examined the relationship between the extent of DNA gain achieved by amplification and the sequence features associated with each amplification strategy, including coverage uniformity, error rate, and utility for producing genome assemblies or finding variants.
Results of the analysis showed an edge for PCR-based approaches such as MALBAC and PicoPLEX when looking at copy number profiles and/or amplifying DNA to high gains, for example. On the other hand, MDA-based amplification appeared to provide an advantage when mapping reads or searching for SNPs, due to its lower per-base error rates.
Informed by these findings, those behind the new comparison plan to stick with that approach for their own upcoming single-cell sequencing projects.
"We have, from the early days, been using MDA," Quake told In Sequence. "But we were interested in testing some of the claims and seeing if we should switch to another application chemistry."
"It depends on the question you're asking," Quake explained. While there is "no single best amplifier," he and his colleagues concluded that the MDA is still the most appropriate for most of their applications.
As researchers pursue single-cell sequencing experiments designed to delve into cellular functions and diversity, methods for preparing sufficient amounts of DNA from ever-smaller samples have become increasingly important.
Generally speaking, the DNA duplication approaches required to achieve this have hinged on amplification methods such as MALBAC and PicoPLEX — which include PCR cycles done at different temperatures — or so-called isothermal approaches such as MDA that rely on highly processive polymerase enzymes prone to DNA strand displacement.
As part of their head-to-head comparison, the researchers assessed 41 sequencing libraries prepared using different combinations of E. coli sample types, WGA approaches, and microfluidic chip amplification chamber volumes.
In either a 50-microliter tube or a 150-nanoliter microfluidic chamber, the team applied MDA to between four and five individual E. coli cells before sequencing DNA from each cell to an average depth of 158-fold coverage with 2X250 sequencing on the Illumina MiSeq.
It also applied a similar sequencing strategy to DNA from bulk E. coli samples or single E. coli cells that had been amplified by MALBAC or PicoPLEX in a test tube.
By aligning the resulting reads to the E. coli reference genome, the researchers were able to assess not only the specificity of the sequences produced from each amplicon, but also the sequence uniformity, amplification biases, and the extent of DNA gains made during the amplification process.
This comparison indicated that "no single method performed best across all criteria and significant differences in characteristics were observed," the study authors noted, adding that "the choice of which amplifier to use will strongly depend on the details of the type of question being asked in any given experiment."
For example, amplifications done in small reaction volumes typically coincide with a greater proportion of correctly mapped reads after amplicon sequencing, the researchers reported. Likewise, reactions with lower amplification gains or fewer steps resulted in relatively high read mapping specificity.
Across the amplification approaches tested, the team found that MDA reactions — particularly those done in the microfluidic chamber — were relatively impervious to contamination. In contrast, low levels of starting DNA amplified by MALBAC or PicoPLEX were more prone to picking up background DNA contaminants, perhaps due to the nature of the steps involved in the PCR protocols.
"We found that the PCR-based chemistries, which require many hands-on steps, are particularly sensitive to failure by contamination," the study's authors wrote.
The team noted that material amplified by MDA tended toward low-frequency noise in coverage uniformity that increased at higher and higher amplification levels, whereas the other two amplification approaches produced higher frequency variation in local genome coverage that was not exacerbated by elevated amplification gains.
Nevertheless, the three amplification approaches showed similar coverage at a fixed sequencing depth of 20-fold coverage since the PCR-based MALBAC and PicoPLEX approaches had better uniformity but more reads that did not map to the reference genome.
The study's authors saw subtle differences in their data when applying the resulting reads to variant detection or de novo genome assembly.
In the case of the MDA-amplified samples, the extent of amplification gain was again important. Consequently, PCR-based methods offered an advantage when the team tried to discern copy number profiles in the samples, in part due to the biases associated with samples amplified to high gain by MDA.
"The two areas where some of these other PCR-based [application approaches] did better were in copy number changes and very large gain applications, where you want to do an enormous amount of amplification," Quake noted. "There are some advantages to those other amplifiers in those situations."
In contrast, the PCR-based methods tended to have a higher per-base error rate, the researchers reported, an effect that seems to reflect the fidelity of the polymerase enzymes used for amplification by MALBAC or PicoPLEX. That error rate difference makes MDA more appealing for those searching for SNPs, they argued, with lower per-base error rates revealing more authentic single nucleotide variants.
Quake noted that the results of his team's comparison may prompt others to take a crack at using higher fidelity polymerase enzymes in PCR-based amplification methods in the future. "Now that we've pointed out some of the issues, it should hopefully stimulate folks to improve the quality" of such methods, he said.
The findings from the WGA analysis are not expected to differ with the use of an alternative high-throughput sequencing platform, though sequencing done for the current comparison relied exclusively on Illumina instruments.
At the moment, the researchers do not have a clear answer on whether the sequence composition, GC-content, and the like can affect the coverage uniformity or amplification biases associated with each of the WGA methods.
There does not appear to be any firm cutoff when considering how much amplification gain is possible before switching from MDA to one of the PCR-based approaches, Quake said. He noted that his team plans to continue applying MDA to amplify DNA in its ongoing experiments aimed at sequencing various microbial and mammalian cell types.