OMICfpp

OMICfpp performs statistical analysis of RNA-Seq paired-design data to detect differential gene expression while accounting for small-sample, high-dimensional challenges and experimental design.

Key Features:

• Case-control binomial testing: Calculates p-values for each case-control pair using a binomial test.
• P-value aggregation (OWA): Aggregates individual p-values via an ordered weighted average (OWA) incorporating a predefined orness parameter.
• Randomization distributions: Generates randomized pairs using between-pairs and complete randomization distributions to create null baselines.
• Randomization p-value: Compares aggregated p-values from observed data with randomized distributions to derive a randomization p-value used as a raw p-value per gene.
• Randomized p-values pattern graphics: Produces randomized p-values pattern graphics to support selection of target genes for experimental validation.
• Benchmarking: Enables comparison of results with edgeR and DESeq2 for paired samples.
• Small-n/high-dimensional support: Employs procedures intended to address situations with small sample sizes relative to large numbers of variables.

Scientific Applications:

• Differential expression in paired RNA-Seq: Detects differentially expressed genes in paired RNA-Seq studies while accounting for experimental pairing.
• Colorectal cancer gene discovery: Applied to public colorectal cancer datasets comprising 68 sample pairs for gene identification and validation.
• Target gene selection for validation: Uses randomized p-values pattern graphics to select candidate genes for experimental validation.
• Method benchmarking and validation: Supports benchmarking against edgeR and DESeq2 and validation via simulations and bibliographic searches.

Methodology:

Calculate per-pair p-values with a binomial test; aggregate individual p-values using an ordered weighted average (OWA) with a predefined orness parameter; generate randomized pairs via between-pairs and complete randomization distributions; compare aggregated observed p-values to randomized distributions to compute randomization p-values; produce randomized p-values pattern graphics.