Fast trimer statistics facilitate accurate decoding of large random DNA barcode sets even at large sequencing error...

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• DNA barcodes are crucial for identifying individual biomolecules in pooled samples.
• Existing DNA error-correcting codes (ECCs) have limitations in handling high error rates and large barcode sets.
• Accurate decoding requires low DNA error rates or robust error-correction methodologies.

Purpose of the Study:

• To develop a DNA error-correction methodology for large-scale biomolecule identification.
• To enable accurate decoding of DNA barcodes with high error rates (up to 20%).
• To investigate the feasibility of using long random DNA barcodes for massive datasets.

Main Methods:

• Utilizing long random DNA barcodes (approx. 34 nt) to tolerate multiple errors (>6).
• Implementing a fast triage method based on trimer occurrence statistics for initial read filtering.
• Leveraging Graphics Processing Units (GPUs) for massive parallel processing to accelerate barcode comparison.

Main Results:

• Achieved 99.9% precision and 98.8% recall with 10% DNA errors for 10^6 barcodes of length 34 nt.
• Demonstrated high precision even at 20% nucleotide error rates.
• Estimated low computational cost ($0.15-$0.60 per million reads) using commodity GPUs.

Conclusions:

• Long random DNA barcodes are effective for high-throughput biomolecule identification with high error tolerance.
• GPU-accelerated trimer-based triage significantly enhances decoding speed and efficiency.
• This method offers a cost-effective solution for large-scale DNA data analysis.