Abstract
Of the amplification methodologies examined in this paper, the multiple displacement amplification products generated the least bias, and produced significantly higher yields of amplified DNA.
Introduction
In fields such as forensics, paleontology, and microbiology, researchers often encounter "minute amounts" of DNA—sometimes sourced from a single cell or a degraded fossil. To sequence this material effectively, Whole Genome Amplification (WGA) is required to generate sufficient mass for sequencing platforms. However, WGA is not a perfect process; it can introduce significant bias, over-amplifying certain genomic regions while leaving others underrepresented. This study, a foundational work from 2006, utilized then-state-of-the-art 454 sequencing to provide a high-resolution map of the systematic biases introduced by various WGA methods.
Methods and Experimental Design
The researchers compared four commercial WGA protocols:
- Multiple Displacement Amplification (MDA): Represented by GenomiPhi and Repli-G, utilizing the highly processive $\phi29$ polymerase under isothermal conditions.
- Primer Extension Preamplification (PEP-PCR): A PCR-based method using random primers.
- Degenerate Oligonucleotide Primed PCR (DOP-PCR): A PCR method using partially degenerate primers.
These methods were tested on two bacterial genomes with contrasting GC content to evaluate performance under different chemical stresses:
- Halobacterium species NRC-1: High GC content (66-68%) and repetitive sequences.
- Campylobacter jejuni: Low GC content (31%) and fewer repeats.
The resulting DNA was sequenced using the 454 Sequencing System. Bias was quantified using the Kolmogorov-Smirnov (KS) test and its associated D-statistic, comparing amplified distributions against unamplified controls at a resolution of 10 to 100 base pairs.
Key Findings: Isothermal Superiority and the GC Barrier
The study revealed that MDA methods significantly outperformed PCR-based alternatives in both yield and coverage.
- Coverage: For the low-GC C. jejuni, MDA achieved approximately 99% genome coverage, whereas DOP-PCR covered only 17%.
- Bias Magnitude: At 100-base resolution, the D-statistics for DOP-PCR were 220 to 252 times higher than the unamplified control. Even the best-performing method, GenomiPhi, showed a 15x to 119x increase in bias depending on the organism.
- Chemical Shift: PCR-based methods altered the measured GC content. DOP-PCR reads were roughly 6% more AT-rich than the actual genome, likely due to primer binding difficulties in high-GC regions.
- Repetitive Regions: Repetitive sequences in Halobacterium acted as "sinks," soaking up amplification resources and leading to extreme over-representation.
Limitations and Platform-Specific Errors
While the study provides a comprehensive look at representation bias, it did not extensively explore sequence fidelity (error rates). A notable complication was the use of the 454 sequencing platform, which is known for errors in "homopolymer" regions (consecutive identical bases).
The researchers noted that some observed errors might be artifacts of the sequencing technology rather than the WGA process itself. Distinguishing between the "photocopier" (WGA) and the "scanner" (sequencer) remains a critical challenge for future genomic studies.
Conclusion
MDA remains the preferred method for whole genome amplification due to its high yield and relatively low bias. However, the study underscores that WGA is never truly neutral, particularly when dealing with high-GC or repetitive genomes. Researchers must remain cognizant of these systematic biases when interpreting data from amplified samples.
Summary updated