SAM

SAM identifies significant genomic features associated with outcomes in RNA-Seq and other sequencing-based comparative genomic experiments using a non-parametric, resampling-based statistical approach.

Key Features:

• Non-parametric resampling for count data: Employs a non-parametric method with resampling to analyze count-based RNA-Seq data without assuming a normal distribution and to accommodate varying sequencing depths.
• Robustness against outliers: Mitigates the influence of anomalous data points that can bias Poisson or negative binomial model-based analyses.
• Versatility across outcome types: Supports analysis of quantitative, survival, two-class, and multiple-class outcomes.
• Empirical comparison to parametric models: Has been compared against Poisson and negative binomial-based methods using simulated and real datasets, demonstrating more consistent pattern discovery.

Scientific Applications:

• Gene and feature discovery in RNA-Seq: Identifies genes and genomic features associated with biological outcomes in RNA-Seq studies.
• Comparative genomic experiments: Enables detection of outcome-associated features across experiments with varying sequencing depths.
• Analysis across outcome types: Applicable to studies involving quantitative traits, survival outcomes, binary (two-class), and multiclass comparisons.

Methodology:

SAM applies a non-parametric resampling technique that accounts for differences in sequencing depths to identify significant features without assuming a specific data distribution.