mdp

mdp computes Molecular Degree of Perturbation (MDP) scores to quantify sample-level transcriptional or molecular deviation from a designated control class in high-dimensional omics datasets.

Key Features:

• Molecular Degree of Perturbation scoring: assigns continuous perturbation scores that quantify the extent of sample-level transcriptional or molecular deviation from a control class.
• Per-gene z-score calculation: computes per-gene z-scores relative to the control distribution.
• Classical and modified z-scores: supports classical z-scores and modified z-scores based on the median absolute deviation (MAD) to provide robustness to outliers and non-normal control distributions.
• Configurable zeroing threshold: applies a configurable z-score zeroing threshold to control which deviations are considered in the perturbation score.
• Aggregation into sample-specific scores: aggregates per-gene z-scores into a single continuous perturbation score for each sample.
• Perturbed-gene identification: identifies genes most strongly perturbed between test and control classes to aid interpretation and biomarker discovery.
• Extension of MDH: builds on the Molecular Distance to Health (MDH) metric of Pankla et al. (2009) and extends it with additional statistical options and improved diagnostics.
• Input data formats: accepts a matrix or data frame containing at least two groups (control and test).

Scientific Applications:

• Sample heterogeneity quantification: measures sample-level heterogeneity in transcriptomic, proteomic, and other high-throughput omics datasets.
• Comparative analysis of test versus control: quantifies molecular deviation of test samples relative to a designated control class in experiments with at least two groups.
• Biomarker discovery and interpretation: facilitates identification and interpretation of genes most strongly perturbed between conditions.

Methodology:

Computes per-gene z-scores relative to the control distribution using classical or MAD-based modified z-scores, applies a configurable z-score zeroing threshold, aggregates per-gene values into sample-specific perturbation scores, and identifies genes most strongly perturbed between test and control classes; methodology builds on the Molecular Distance to Health (MDH) metric (Pankla et al., 2009).