Sequencing Saturation Does Not Uniquely Determine Molecular Recovery in UMI Transcriptomics

Area of Science:

• * Molecular biology
• * Bioinformatics
• * Genomics

Background:

• * Current UMI-based transcriptomic experiments rely on heuristic metrics like reads per cell and sequencing saturation for depth planning.
• * Sequencing saturation alone is insufficient for determining molecular recovery due to amplification heterogeneity.

Purpose of the Study:

• * To introduce NB-Lib, a novel modeling framework for UMI-based transcriptomic experiments.
• * To accurately estimate amplification heterogeneity and library complexity.
• * To establish a unified framework for interpreting sequencing saturation, molecular recovery, and sequencing efficiency.

Main Methods:

• * Developed NB-Lib, a modeling framework using a zero-truncated negative binomial representation.
• * Applied NB-Lib to 150 single-cell and spatial transcriptomic datasets.
• * Jointly estimated amplification heterogeneity and library complexity.

Main Results:

• * NB-Lib accurately reconstructs sequencing saturation curves and predicts sequencing depth requirements.
• * Amplification heterogeneity and library complexity provide a compact parameterization linking sequencing depth, saturation, and molecular recovery.
• * Molecular recovery was shown to directly impact the reproducibility of low-abundance gene detection.

Conclusions:

• * NB-Lib offers a unified framework for UMI-based transcriptomic technologies.
• * The framework enables accurate sequencing depth planning and improved molecular recovery.
• * This approach enhances the reproducibility of gene detection in transcriptomic studies.