Using expected sequence features to improve basecalling accuracy of amplicon pyrosequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Amplicon pyrosequencing relies on known genetic regions, expecting specific sequence features like open reading frames.
• Pyrosequencing errors, primarily indels, can disrupt these expected features, including open reading frames.
• Prior knowledge of expected sequence features can guide basecalling, the process of inferring nucleotide sequences from raw data.

Purpose of the Study:

• To develop and evaluate a novel basecalling method, Multipass, for amplicon pyrosequencing data.
• To improve the accuracy and reliability of nucleotide sequence inference by integrating prior biological knowledge into the basecalling process.

Main Methods:

• Multipass employs a probabilistic framework to analyze raw pyrosequencing flowgram data.
• It calculates the likelihood of multiple potential nucleotide sequences for each variant.
• The method integrates basecalling with models of expected sequence characteristics, such as open reading frames.

Main Results:

• Multipass demonstrated a 20% increase in error-free sequences compared to existing state-of-the-art methods on 454 amplicon pyrosequencing data.
• The method successfully generated high-confidence, error-free sequences from a malaria virulence gene family dataset.
• The probabilistic approach allows for the incorporation of additional parameters, enhancing basecalling accuracy.

Conclusions:

• The Multipass method enhances the accuracy of amplicon sequencing data, particularly when prior knowledge about target sequences is available.
• It is applicable to existing and future sequencing data, with potential applications in areas like immunoglobulin VDJ region analysis.
• Multipass is available for Roche 454 data, with the potential for implementation on other sequencing platforms.