A computational method for estimating the PCR duplication rate in DNA and RNA-seq experiments.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Polymerase chain reaction (PCR) amplification is crucial for DNA sequencing library preparation.
• PCR amplification can introduce redundant reads, necessitating accurate estimation of PCR duplication rates.
• Existing methods overestimate PCR duplication by failing to differentiate between PCR duplicates and natural read duplicates.

Purpose of the Study:

• To develop a computational method for accurately estimating average PCR duplication rates in high-throughput sequencing datasets.
• To differentiate PCR duplicates from natural read duplicates in DNA-seq and RNA-seq experiments.
• To improve the assessment of read duplication frequency in sequencing data.

Main Methods:

• Developed a novel computational method leveraging heterozygous variants to distinguish PCR duplicates from natural duplicates.
• Applied the method to simulated data and exome sequence data from the 1000 Genomes project.
• Validated the method on paired-end and single-end read datasets, including those with high proportions of natural duplicates.

Main Results:

• The method accurately estimates PCR duplication rates in both simulated and real-world exome and RNA-seq datasets.
• Analysis revealed that 45-50% of read duplicates in exome data (Nextera preparation) are natural duplicates, attributed to fragmentation bias.
• In RNA-seq data, 70-95% of observed read duplicates are natural duplicates from highly expressed genes, with outlier samples showing doubled PCR duplication rates.

Conclusions:

• The developed method provides a valuable tool for estimating PCR duplication rates and quantifying natural read duplicates in high-throughput sequencing data.
• This approach enhances the reliability of sequence data analysis by accurately accounting for different types of read duplication.
• An implementation of the method is publicly available for researchers.