Correcting 4sU induced quantification bias in nucleotide conversion RNA-seq data.

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Nucleoside analogues like 4-thiouridine (4sU) enable metabolic labeling of newly synthesized RNA.
• Chemical conversion of 4sU to C:T mismatches allows simultaneous profiling of total and labeled RNA expression.
• Potential cytotoxicity and impact of extensive 4sU labeling on RNA-seq expression estimates remain understudied.

Purpose of the Study:

• To investigate the effects of escalating 4-thiouridine (4sU) doses and prolonged labeling times on expression estimates in nucleotide conversion RNA-sequencing.
• To identify the mechanisms underlying expression bias induced by extensive 4sU labeling.
• To develop computational and statistical methods for correcting these biases.

Main Methods:

• Nucleotide conversion RNA-sequencing was performed using varying concentrations and time courses of 4sU labeling across different cell lines.
• Analysis focused on identifying biases in expression estimates related to RNA half-life.
• Development of a computational tool for rescuing unmappable reads and a statistical method for bias correction.

Main Results:

• Expression estimates were biased in an RNA half-life dependent manner at high 4sU concentrations or later time points.
• Bias was attributed to reduced read mappability with multiple conversions and underrepresentation of labeled RNA during library preparation.
• A novel computational tool improved read mapping, and a statistical method effectively removed remaining bias.

Conclusions:

• Extensive 4-thiouridine labeling can introduce significant bias into RNA sequencing expression estimates.
• Developed computational and statistical methods, integrated into the GRAND-SLAM pipeline and grandR package, can accurately correct these biases.
• These advancements enable more reliable quantification of newly synthesized RNA using nucleotide conversion RNA-seq.