NBGLM-LBC

NBGLM-LBC corrects gene-specific library biases in highly multiplexed RNA sequencing studies to enable accurate integration of bulk RNA and single-cell RNAseq data that use DNA barcodes and pooled barcoded cDNAs.

Key Features:

• Bias Correction Capability: Corrects biases arising during preparation of multiple library pools that can produce low correlation between expression profiles and introduce batch-effect biases.
• Algorithmic Approach: Implements a non-linear generalized linear model (NBGLM) specifically tailored to correct gene-specific library biases across RNAseq libraries.
• Handling Uneven Library Yields: Recommends omission of libraries with particularly low yields as a strategy to manage uneven library yields and improve data integration.
• Simulation and Application: Uses simulation experiments to show that effective bias correction requires a consistent sample layout and was validated in a childhood acute respiratory illness cohort where library biases were resolved.
• Applicability: Applicable to highly multiplexed sequencing-based profiling methods that use DNA barcodes and pooled barcoded cDNAs with equally distributed samples (e.g., "cases" and "controls") per library.

Scientific Applications:

• Large-scale transcriptome studies: Improves reliability of expression measurements in studies employing multiplexed RNAseq, including bulk RNA and single-cell RNAseq.
• Low-input and single-cell profiling: Enhances comparability of low-input RNAseq and single-cell RNAseq datasets by correcting library-specific biases.
• Comparative and differential analysis: Enables more accurate sample-to-sample comparisons and downstream differential expression analyses by removing library-induced distortions.
• Clinical cohort studies: Supports analysis of clinical cohorts investigating disease mechanisms and potential therapeutic targets, as demonstrated in a childhood acute respiratory illness cohort.

Methodology:

Applies a non-linear generalized linear model (NBGLM) for gene-specific bias correction and evaluates performance using simulation experiments.