Limits...
Testing the reproducibility of multiple displacement amplification on genomes of clonal endosymbiont populations.

Ellegaard KM, Klasson L, Andersson SG - PLoS ONE (2013)

Bottom Line: The amplified DNA was sequenced with the Illumina sequencing technology, and the results were compared to sequence data obtained from unamplified DNA in this study as well as from a previously published genome project.We conclude that many of the artifacts associated with the use of the multiple displacement amplification method can be alleviated or much reduced by using multiple cells as the template for the amplification.These findings should be particularly useful for researchers studying the genomes of endosymbionts and other uncultured bacteria, for which a small clonal population of cells can be isolated.

View Article: PubMed Central - PubMed

Affiliation: Department of Molecular Evolution, Cell and Molecular Biology, Science for Life Laboratory, Biomedical Centre, Uppsala University, Uppsala, Sweden.

ABSTRACT
The multiple displacement amplification method has revolutionized genomic studies of uncultured bacteria, where the extraction of pure DNA in sufficient quantity for next-generation sequencing is challenging. However, the method is problematic in that it amplifies the target DNA unevenly, induces the formation of chimeric reads and also amplifies contaminating DNA. Here, we have tested the reproducibility of the multiple displacement amplification method using serial dilutions of extracted genomic DNA and intact cells from the cultured endosymbiont Bartonella australis. The amplified DNA was sequenced with the Illumina sequencing technology, and the results were compared to sequence data obtained from unamplified DNA in this study as well as from a previously published genome project. We show that artifacts such as the extent of the amplification bias, the percentage of chimeric reads and the relative fraction of contaminating DNA increase dramatically for the smallest amounts of template DNA. The pattern of read coverage was reproducibly obtained for samples with higher amounts of template DNA, suggesting that the bias is non-random and genome-specific. A re-analysis of previously published sequence data obtained after amplification from clonal endosymbiont populations confirmed these predictions. We conclude that many of the artifacts associated with the use of the multiple displacement amplification method can be alleviated or much reduced by using multiple cells as the template for the amplification. These findings should be particularly useful for researchers studying the genomes of endosymbionts and other uncultured bacteria, for which a small clonal population of cells can be isolated.

Show MeSH
Coverage correlation plots.Pairwise comparisons of the coverage of sequence reads (averaged in 1 kb windows) across the B. australis genome in the Illumina data sets of MDA samples taken from a dilution series of intact bacterial cells, as detailed in Table 1. (A) "cells2" versus "control 1", (B) "cells2" versus "cells3", (C) "cells2" versus "cells4", (D) "cells2" versus "cells5". Sample "control 1" refers to the unamplified Illumina data set obtained from the re-sequencing of the B. australis in this study.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3842359&req=5

pone-0082319-g004: Coverage correlation plots.Pairwise comparisons of the coverage of sequence reads (averaged in 1 kb windows) across the B. australis genome in the Illumina data sets of MDA samples taken from a dilution series of intact bacterial cells, as detailed in Table 1. (A) "cells2" versus "control 1", (B) "cells2" versus "cells3", (C) "cells2" versus "cells4", (D) "cells2" versus "cells5". Sample "control 1" refers to the unamplified Illumina data set obtained from the re-sequencing of the B. australis in this study.

Mentions: For the MDA samples with the lowest amount of template DNA, such as “cells5” and “gDNA8”, there was no correlation to any of the other samples despite the co-localization of several coverage peaks. Even so, when plotting the coverage data between datasets obtained from high and low amounts of template DNA, such as for example from "cells2" and "cells5", we observed that high-coverage regions in sample "cells5" were also over-represented in sample "cells2" (Figure 4). However, some of the highest coverage regions in sample "cells2" contained no reads in sample "cells5", which explains the overall lack of correlation. This indicates that the pattern of the amplification is similar in all samples, but gets increasingly noisy with lower amounts of template DNA, presumably because of stochastic variations in the order in which the primers bind to the few molecules of template DNA available.


Testing the reproducibility of multiple displacement amplification on genomes of clonal endosymbiont populations.

Ellegaard KM, Klasson L, Andersson SG - PLoS ONE (2013)

Coverage correlation plots.Pairwise comparisons of the coverage of sequence reads (averaged in 1 kb windows) across the B. australis genome in the Illumina data sets of MDA samples taken from a dilution series of intact bacterial cells, as detailed in Table 1. (A) "cells2" versus "control 1", (B) "cells2" versus "cells3", (C) "cells2" versus "cells4", (D) "cells2" versus "cells5". Sample "control 1" refers to the unamplified Illumina data set obtained from the re-sequencing of the B. australis in this study.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3842359&req=5

pone-0082319-g004: Coverage correlation plots.Pairwise comparisons of the coverage of sequence reads (averaged in 1 kb windows) across the B. australis genome in the Illumina data sets of MDA samples taken from a dilution series of intact bacterial cells, as detailed in Table 1. (A) "cells2" versus "control 1", (B) "cells2" versus "cells3", (C) "cells2" versus "cells4", (D) "cells2" versus "cells5". Sample "control 1" refers to the unamplified Illumina data set obtained from the re-sequencing of the B. australis in this study.
Mentions: For the MDA samples with the lowest amount of template DNA, such as “cells5” and “gDNA8”, there was no correlation to any of the other samples despite the co-localization of several coverage peaks. Even so, when plotting the coverage data between datasets obtained from high and low amounts of template DNA, such as for example from "cells2" and "cells5", we observed that high-coverage regions in sample "cells5" were also over-represented in sample "cells2" (Figure 4). However, some of the highest coverage regions in sample "cells2" contained no reads in sample "cells5", which explains the overall lack of correlation. This indicates that the pattern of the amplification is similar in all samples, but gets increasingly noisy with lower amounts of template DNA, presumably because of stochastic variations in the order in which the primers bind to the few molecules of template DNA available.

Bottom Line: The amplified DNA was sequenced with the Illumina sequencing technology, and the results were compared to sequence data obtained from unamplified DNA in this study as well as from a previously published genome project.We conclude that many of the artifacts associated with the use of the multiple displacement amplification method can be alleviated or much reduced by using multiple cells as the template for the amplification.These findings should be particularly useful for researchers studying the genomes of endosymbionts and other uncultured bacteria, for which a small clonal population of cells can be isolated.

View Article: PubMed Central - PubMed

Affiliation: Department of Molecular Evolution, Cell and Molecular Biology, Science for Life Laboratory, Biomedical Centre, Uppsala University, Uppsala, Sweden.

ABSTRACT
The multiple displacement amplification method has revolutionized genomic studies of uncultured bacteria, where the extraction of pure DNA in sufficient quantity for next-generation sequencing is challenging. However, the method is problematic in that it amplifies the target DNA unevenly, induces the formation of chimeric reads and also amplifies contaminating DNA. Here, we have tested the reproducibility of the multiple displacement amplification method using serial dilutions of extracted genomic DNA and intact cells from the cultured endosymbiont Bartonella australis. The amplified DNA was sequenced with the Illumina sequencing technology, and the results were compared to sequence data obtained from unamplified DNA in this study as well as from a previously published genome project. We show that artifacts such as the extent of the amplification bias, the percentage of chimeric reads and the relative fraction of contaminating DNA increase dramatically for the smallest amounts of template DNA. The pattern of read coverage was reproducibly obtained for samples with higher amounts of template DNA, suggesting that the bias is non-random and genome-specific. A re-analysis of previously published sequence data obtained after amplification from clonal endosymbiont populations confirmed these predictions. We conclude that many of the artifacts associated with the use of the multiple displacement amplification method can be alleviated or much reduced by using multiple cells as the template for the amplification. These findings should be particularly useful for researchers studying the genomes of endosymbionts and other uncultured bacteria, for which a small clonal population of cells can be isolated.

Show MeSH