Analyze R Package

analyze-r-package

Analyze an R/Bioconductor package to understand its structure, purpose, and key characteristics.

Usage

Invoke this skill when you want to understand an R package's architecture:

• "Analyze this R package"
• "Tell me about this package structure"

Prerequisites

• Working directory is an R package root (contains DESCRIPTION file)
• Package has standard R structure (R/, NAMESPACE, etc.)

Process

1. Read Package Metadata: Analyze DESCRIPTION for name, purpose, version, dependencies, and classify its type (Data, Analysis, Infrastructure, Utility) based on biocViews.
2. Identify Exports: Parse NAMESPACE for exported functions and (if present) S4 classes/methods. Categorize exports broadly (Data access, Processing, Visualization, Utility).
3. Examine Directory Structure: Note the presence of data/, inst/extdata/, vignettes/, tests/testthat/, and src/.
4. Detect Data Access Patterns: Search R/ source files for remote data access (ExperimentHub, AnnotationHub, DuckDB, AWS S3, HuggingFace, etc.) vs local data access.
5. Identify Classes: List R class definitions (S3, S4, R6, S7) and their properties/slots/fields.
6. Read README: Extract high-level purpose and key features.
7. Analyze Testing: Check tests/testthat/ structure, file count, and remote vs local test data.
8. List Vignettes: Check vignettes/ for .Rmd files, extracting titles and purposes.

Output Format

Produce a structured markdown summary:

## Package Analysis: [Package Name]

### Classification
- **Type**: [Data/Analysis/Infrastructure/Utility]
- **Purpose**: [1-2 sentence summary]
- **Version**: [version number]

### Key Exports ([count] total)
- **Data Access Functions**: [list]
- **Data Processing Functions**: [list]
- **Utility Functions**: [list]

### Data Access Pattern
- **Type**: [None / Local Only / Remote / Hybrid]
- **Technologies**: [e.g., ExperimentHub, DuckDB]

### Classes
- [ClassName] - [Type: S3/S4/R6/S7] - [description]

### Documentation & Testing
- **Vignettes**: [List with titles]
- **Testing**: [Framework, count, data types]

### Special Characteristics
[List notable patterns that should be documented]

### Dependencies of Note
[List key Bioconductor or specialized packages]

Examples

Example: Data Package Analysis

User: "Analyze this R package"

Agent: (Analyzes parkinsonsMetagenomicData repository)

## Package Analysis: parkinsonsMetagenomicData

### Classification
- **Type**: Data Package
- **Purpose**: Provides uniformly processed gut microbiome data via remote parquet files accessed through DuckDB.
- **Version**: 0.99.0

### Key Exports (18 total)
**Data Access Functions** (5):
- `returnSamples()` - Main high-level data retrieval function
- `loadParquetData()` - Load filtered data from DuckDB connection

**Discovery Functions** (5):
- `parquet_colinfo()` - Inspect column structure
- `biobakery_files()` - List available data types

### Data Access Pattern
- **Type**: Hybrid (Remote primary, Local for testing)
- **Technologies**: DuckDB for remote parquet access, TreeSummarizedExperiment output

### Documentation & Testing
**Vignettes** (4):
1. codebook.Rmd - Data Codebook
2. full-workflow.Rmd - Comprehensive tutorial

**Testing**:
- Framework: testthat (3 files)
- Test data: inst/extdata/ (parquet, TSV, RDS)

### Special Characteristics
- Uses DuckDB for efficient remote parquet file querying without full download

Integration

This analysis output is consumed by create-package-instructions and update-package-instructions.

See also: create-package-instructions