mandalaR

Comprehensive documentation for Mandala, an R mixed-model framework for plant breeding and quantitative genetics.

About Mandala

Mandala is being developed as a breeder-oriented statistical framework for analyzing plant breeding experiments. Its primary focus is mixed-model analysis of field trials, multi-environment trials, genomic prediction, spatial models, and prediction of genetic values from complex breeding data.

Mandala is written for the R ecosystem with a deliberately small dependency footprint. The package aims to keep the user-facing workflow transparent and portable while using established mixed-model methodology, including REML, mixed-model equations, AI-REML updates, sparse matrix computation, and prediction from fitted linear mixed models.

The goal is to provide practical R syntax for common breeding workflows while retaining access to modern variance structures and efficient sparse mixed-model computation.

Statistical Scope

Mandala currently emphasizes:

• single-site and multi-environment trial analysis
• genotype BLUPs and genotype BLUEs
• genomic relationship models and GBLUP
• sparse and partially replicated MET designs
• spatial row-column and residual covariance models
• single-stage and two-stage analysis workflows
• factor-analytic and other advanced variance structures

Multivariate modeling is under development, but is not the main focus of the current public documentation.

Current Development Status

Mandala is an active development package. The current implementation includes an MME-based AI-REML engine, selected prediction standard errors for large models, selected fixed-effect tests, genomic prediction utilities, spatial diagnostics, and stage-wise analysis tools.

The package is still being refined for public release. Documentation in this repository should therefore be viewed as a development preview rather than a final release manual.

Installation instructions will be added when a public binary or approved distribution channel is available.

Documentation

Rendered documentation should be viewed through the GitHub Pages site:

https://brbasnet.github.io/mandalaR/

Direct links:

• Mandala documentation home
• Introduction to Mandala
• Single-Stage to Multi-Stage Trial Analysis
• Genomic Prediction
• Spatial Analysis
• Advanced Variance Structures
• Mandala Package Comparison

Note: clicking .html files inside the GitHub repository file browser shows the HTML source code. Use the GitHub Pages links above to view rendered pages.

Public Example Data

The small datasets used in the vignettes are included in the data/ folder so that examples can be reproduced by users and learners.

• fullrep_MET_n1000.csv: full-replicated MET example
• sparse_prep_MET_n1000.csv: sparse/partially replicated MET example
• augmented_single_n200.csv: augmented single-site spatial example
• sim_GRM_1000.rds: genomic relationship matrix
• Dataset notes

Repository Scope

This repository is intended for public documentation and communication. It may include:

• rendered vignettes
• public examples
• documentation pages
• benchmark summaries
• package-positioning material
• roadmap notes

It should not include:

• unreleased implementation scripts
• internal development archives

Additional project information will be shared through Listo Agriculture.

Local Preview

Open the documentation home page directly:

open index.html

Or serve locally:

python3 -m http.server 8000

Then open:

http://localhost:8000