Shining a light on dark sequencing: characterising errors in Ion Torrent PGM data.

Area of Science:

• Genomics
• Next-Generation Sequencing
• Bioinformatics

Background:

• The Ion Torrent Personal Genome Machine (PGM) utilizes pH measurement for sequencing, differing from light-based methods.
• Understanding platform-specific errors is crucial for accurate genomic data interpretation.

Purpose of the Study:

• To comprehensively characterize biases and errors in the Ion Torrent PGM sequencing platform.
• To identify the main sources and types of errors, including insertion/deletion (indel) errors.

Main Methods:

• Analysis of re-sequencing datasets generated by the PGM.
• Evaluation of error rates across various factors: chip density, sequencing kit, template species, and machine.
• Characterization of two distinct indel error types: inaccurate flow-calls and high-frequency indels (HFI).

Main Results:

• Inaccurate flow-calls were the primary error source, causing indels at a raw rate of 2.84% (1.38% after quality clipping).
• Flow-call errors led to over-calling short homopolymers and under-calling long homopolymers, with accuracy decreasing over flow cycles.
• High-frequency indel (HFI) errors occurred less frequently (0.06% of bases) but were inconsistent across runs.

Conclusions:

• The PGM currently exhibits lower accuracy than competing light-based technologies.
• Systematic flow-call inaccuracies can be addressed by developing PGM-specific bioinformatics models.
• HFI errors present challenges for polymorphism and amplicon studies but may be mitigated by multi-chip sequencing.