EnsMOD

EnsMOD detects outliers in omics datasets to identify aberrant samples and improve the integrity of downstream biological and statistical analyses.

Key Features:

• Transcriptomic outlier algorithms: Integrates two robust algorithms specifically developed for transcriptomic outlier detection.
• Quantitation Variation Analysis: Evaluates how closely quantitation variation in samples adheres to a normal distribution.
• Density Curve Visualization: Plots density curves for each sample to reveal anomalies indicative of outliers.
• Hierarchical Cluster Analysis: Assesses sample clustering to identify samples that deviate from expected groups.
• Robust Principal Component Analysis (PCA): Employs robust PCA to statistically test whether any sample deviates significantly from others.
• Adjustable Probabilistic Thresholds: Provides probabilistic threshold parameters to modify the stringency of outlier detection.
• Versatility Across Omics Datasets: Applicable to datasets with normally distributed variance, including simulated proteomics, multiomic (proteome and transcriptome), single-cell proteomics, and phosphoproteomics.

Scientific Applications:

• Omics data quality control: Identifies and facilitates removal of sample outliers to improve the reliability of downstream statistical analyses.
• Proteomics and phosphoproteomics: Applied to simulated proteomics and phosphoproteomics datasets for outlier identification.
• Multiomic datasets: Applied to proteome and transcriptome multiomic datasets to assess sample integrity across modalities.
• Single-cell proteomics: Applied to single-cell proteomics datasets to detect aberrant single-cell samples.

Methodology:

Applies two transcriptomic outlier-detection algorithms, quantitation variation analysis (assessing adherence to a normal distribution), density curve plotting per sample, hierarchical clustering, robust principal component analysis (PCA), and probabilistic thresholding.