Illumina error profiles: resolving fine-scale variation in metagenomic sequencing data

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Illumina sequencing platforms are widely used globally, offering high throughput and low costs.
• Rapid advancements in sequencing technology have outpaced the understanding of associated data errors and biases.
• Noise in sequencing data necessitates a deeper investigation into platform-specific error profiles.

Purpose of the Study:

• To systematically investigate errors and biases in Illumina sequencing data.
• To evaluate different Illumina sequencing platforms (Genome Analyzer II, HiSeq, MiSeq) and library preparation methods.
• To identify sequence motifs and nucleotide incorporation biases specific to the sequencing-by-synthesis process.

Main Methods:

• Analysis of the largest collection of in vitro metagenomic datasets to date.
• Position- and nucleotide-specific analysis of sequencing errors.
• Evaluation of quality-score-based error removal strategies.

Main Results:

• A significant bias was identified in motifs (3mers) preceding errors, particularly those ending in 'GG', accounting for approximately 16% of substitution errors.
• Preferential incorporation of ddGTPs was observed, likely due to the engineered polymerase and ddNTPs in sequencing-by-synthesis.
• Quality-score-based error removal eliminated about 69% of substitution errors, but the identified motif bias persisted.

Conclusions:

• Detecting single-nucleotide polymorphism changes in bacterial genomes is vital for understanding phenotypes like antibiotic resistance and virulence.
• Existing error removal techniques are insufficient for addressing Illumina-specific biases, potentially impacting downstream analyses and conclusions.
• Distinguishing systematic sequencing errors from true genetic variation is essential for developing accurate diagnostic and therapeutic strategies.