RiskyCNV: RiskyCNV: A General-Purpose CNV Analysis Workflow for Disease...
In RiskyCNV: Risk Analysis of Genomic Copy Number Variation

RiskyCNV

R Documentation

RiskyCNV: A General-Purpose CNV Analysis Workflow for Disease Risk Stratification

Description

Provides a complete seven-step workflow for copy number variation (CNV) analysis applicable to any disease or condition where genomic copy number data is available. The package supports three classification approaches: built-in grading and risk stratification presets for seven major disease types (prostate, breast, colorectal, lung, cervical, lymphoma, melanoma) based on clinically validated scoring systems; an automatic mode that derives a normalised Risk Score from the data itself using min-max normalisation and adaptive binning; and fully user-defined custom thresholds for any disease type not covered by the presets. Downstream functions for CNV aberration detection, recurrence analysis, gene annotation, CNV matrix generation, and CNV-RNA expression correlation are disease-type agnostic and work with any genomic dataset.

Workflow

The recommended analysis pipeline proceeds in seven steps:

extract_metadata — Classify samples into grade or stage groups based on a clinical scoring parameter. Supports disease-specific presets, auto mode, and custom thresholds.
classify_risk — Stratify samples into risk categories. Supports disease-specific presets, auto mode, and custom thresholds.
aberration — Detect CNV gains and losses from segmented CNV data using a user-defined effect size threshold.
recurrent — Identify CNV regions that recur across multiple samples within a risk group.
annotate — Annotate recurrent CNV regions with gene symbols using genomic range overlaps.
create_CNVMatrix — Construct a sample x gene CNV expression matrix.
correlate_with_expr — Compute Pearson correlations between CNV profiles and RNA expression data.

Classification Modes

Disease presets: Built-in clinically validated thresholds for prostate (ISUP/Gleason), breast (Nottingham), colorectal (Dukes), lung (IASLC/TNM), cervical (FIGO), lymphoma (Ann Arbor/Lugano), and melanoma (Breslow depth).
Auto mode: Automatically computes a normalised Risk Score using min-max normalisation. Group boundaries are determined adaptively based on distribution skewness — equal-width for symmetric distributions, quantile-based for skewed distributions. Works for any numeric scoring column without prior knowledge of the scoring system.
Custom mode: Users supply their own threshold functions for any disease type, scoring system, or number of groups.

Supported Disease Types (Built-in Presets)

prostate (ISUP/Gleason), breast (Nottingham/NGS), colorectal (Dukes), lung (IASLC/TNM), cervical (FIGO), lymphoma (Ann Arbor/Lugano), and melanoma (Breslow depth).