TRUmiCount: correctly counting absolute numbers of molecules using unique molecular identifiers

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Quantitative next-generation sequencing (NGS) methods like RNA-Seq are limited by PCR amplification bias.
• Unique molecular identifiers (UMIs) help distinguish molecules but cannot fully correct for PCR bias or sequencing losses.
• Existing methods struggle with both underestimation due to molecule loss and overestimation from amplification artifacts like phantom UMIs.

Purpose of the Study:

• To introduce the TRUmiCount algorithm for accurate molecule counting in NGS.
• To develop a method that corrects for PCR amplification bias and sequencing losses.
• To provide a robust quantification tool applicable to various NGS applications, including single-cell RNA-Seq.

Main Methods:

• TRUmiCount employs a mechanistic model of PCR amplification and sequencing.
• The algorithm estimates PCR efficiency and sequencing depth directly from experimental data, eliminating the need for calibration experiments or spike-ins.
• It filters out phantom UMIs and estimates molecules lost during sequencing.

Main Results:

• TRUmiCount effectively corrects for both phantom UMIs and molecule loss during sequencing.
• The algorithm's parameters (PCR efficiency, sequencing depth) are physically interpretable and estimable from data.
• Phantom-filtered and loss-corrected molecule counts show significantly higher accuracy compared to raw UMI counts.

Conclusions:

• TRUmiCount provides a more accurate method for quantifying molecules in NGS experiments.
• The algorithm addresses key sources of error in UMI-based quantification.
• TRUmiCount is a valuable tool for improving the reliability of quantitative NGS data, particularly in single-cell applications.