aFold

aFold estimates differential gene expression from RNA sequencing (RNA-Seq) data by applying qtotal normalization and polynomial uncertainty modelling to produce fold-change values comparable across experiments.

Key Features:

• qtotal normalization: qtotal accounts for the overall distribution of read counts to standardize RNA-Seq data and mitigate effects of asymmetrical up- and down-regulation and outliers.
• Polynomial uncertainty modelling: Polynomial algorithms model and quantify uncertainty in read counts to improve the precision of differential expression estimates across treatments and genes.
• Benchmarking and validation: Benchmarking on simulated and validated real-life datasets, including ABRF, SEQC, and MAQC-II, demonstrates improved identification of differentially expressed genes, particularly under asymmetrical distributions and in the presence of outliers.
• Facilitation of downstream analyses: Infers fold-change values that are comparable across experiments to support data clustering, visualization, and other downstream analyses.

Scientific Applications:

• Data normalization: Standardizes RNA-Seq datasets using qtotal when traditional assumptions like symmetrical gene expression distributions do not hold.
• Differential expression analysis: Detects genes with significant expression changes across experimental conditions using polynomial uncertainty modelling.
• Handling real-life data complexity: Manages noise, asymmetrical distributions, and outliers in RNA-Seq data to increase robustness of differential expression results.

Methodology:

Two explicit computational steps: normalization via qtotal to adjust read counts for overall distribution patterns, and polynomial modelling of uncertainty to quantify variability in read counts for differential expression estimation.