mintyr

mintyr is a high-performance data processing toolkit designed specifically for animal breeding and genomic selection. Leveraging the zero-copy and multi-threading capabilities of data.table, it significantly simplifies the construction of automated data pipelines in large-scale commercial breeding programs (e.g., coordinating data across nucleus and multiplier farms, or handling multi-trait growth test records).

The package is not only highly optimized for iterative analysis workflows with the ASReml-R package (supporting dynamic modeling and multi-trait/multi-breed nested grouping), but is also capable of generating and batch-exporting formatted phenotypic data files required for automated pipeline analyses in other mainstream command-line breeding software (e.g., HIBLUP, DMU).

✨ Key Features

mintyr covers five critical stages in the lifecycle of breeding data analysis:

🚀 High-Performance Data I/O (import_xlsx, import_csv, export_xlsx)

A transparent round-trip for multi-file, multi-sheet tabular data: import many files into one tidy data.table, transform freely, then write the original file/sheet structure back out — no bookkeeping required.

• Import (import_xlsx, import_csv): native support for merging multiple files and sheets simultaneously, with source tracking columns (excel_name, sheet_name) appended automatically to prevent data confusion across different farms or batches. In-place data.table conversion keeps the memory footprint minimal, and import_xlsx can spread the per-sheet parse across CPU cores on demand (opt-in via workers) — a fork pool on Linux/macOS, a PSOCK cluster on Windows.
• Export (export_xlsx): the round-trip companion — a single path argument decides the destination. A directory writes one .xlsx per excel_name value (one sheet per sheet_name); a .xlsx file path writes everything into one workbook. Worksheet splitting follows the data automatically, and the tracking columns are stripped by default so exported sheets match the originals.

🔄 Automated Data Reshaping & Nesting (w2l_nest, c2p_nest, r2p_nest)

• c2p_nest: Column-to-pairs nested transformation that automatically renames feature columns, providing standard uniform inputs for iterative multi-trait genetic correlation evaluations.
• w2l_nest / w2l_split: Wide-to-long format transformations with subsetting and nesting by grouping variables (e.g., farm, breed, or line).

🧪 Cross-Validation & Model Evaluation (split_cv, nest_cv)

• Provides streamlined k-fold cross-validation and repeated CV splitting tailored for nested data.table structures, facilitating the evaluation of breeding value prediction accuracy (GP).

📊 Batch Exporting for Breeding Software (export_nest, export_list)

• Automatically builds local folder hierarchies and batch-exports data based on nested groups or list structures (e.g., tempdir()/Line/Breed/data.txt), providing seamless text-file preparation to bridge the gap with command-line driven breeding evaluation software like HIBLUP and DMU.

🛠️ Phenotypic Statistics & Preprocessing (top_perc, format_digits, get_path_info)

• Quickly extracts the top N% performing individuals for specific traits, supporting independent extraction by analytical groups (e.g., different test batches), alongside robust number formatting and path-parsing utilities.

📥 Installation

You can install this package from either CRAN or GitHub:

### From CRAN
install.packages("mintyr")
### From GitHub
pak::pak("tony2015116/mintyr")

📋️ Cheat Sheet

🤝 Acknowledgments

Special thanks to AI assistance, for helping transform the initial concepts and inspirations for the mintyr package into reality. Their contribution has been invaluable in refining ideas, improving code structure, and crafting documentation.