SimGP: Simulation Framework for Genomic Prediction

SimGP is an open-source, R-based stochastic simulation framework for generating biologically realistic breeding populations and analysis-ready datasets for benchmarking genomic prediction methods.

This repository contains SimGP version 1.0.0: the tomato case-study implementation used in the accompanying manuscript. Tomato serves as the first demonstration of the framework. The simulation parameters can be readily modified to explore alternative population sizes, genome structures, trait architectures, field designs and validation scenarios.

---

Key features

• Open-source R framework for benchmarking genomic prediction methods.
• Generates biologically realistic breeding populations.
• Simulates genome-wide SNP markers and causal QTL.
• Supports both trait-specific and pleiotropic genetic architectures.
• Simulates true breeding values for multiple quantitative traits.
• Models genotype-by-environment interaction together with environmental, replicate, block, spatial and residual effects.
• Produces analysis-ready phenotype, genotype and validation datasets.
• Easily adapted to alternative species and breeding scenarios by modifying the configuration parameters.

---

What the tomato implementation simulates

• 20 biparental families and 200 F6 genotypes by default
• 2,580 SNP markers across 12 tomato chromosomes
• trait-specific and pleiotropic causal QTLs
• true breeding values for five traits
• genotype-by-environment interaction
• environmental, replicate, block, spatial and residual effects
• random missing phenotypic observations
• five-fold and family-based validation assignments
• marker- and genotype-level quality-control summaries

The simulated traits are:

• Yield
• Fruit weight
• Brix
• Days to flowering
• Bacterial wilt severity

The bacterial wilt disease score ranges from 1 (highly resistant) to 9 (highly susceptible).

---

Requirements

SimGP v1.0.0 uses base R only.

Recommended R version: R 4.1 or later.

---

Quick start

Clone or download the repository, set the repository as the working directory in R, and run:

source("SimGP_tomato_case_study.R")

The default output directory is

SimGP_tomato_case_study_output/

---

Simulation workflow

Configuration
    |
    v
Population simulation
    |
    v
Marker simulation
    |
    v
QTL and true breeding value simulation
    |
    v
Field trial simulation
    |
    v
Phenotype simulation
    |
    v
Validation dataset creation
    |
    v
Quality-control summaries
    |
    v
CSV export

---

Modify simulation parameters

Simulation parameters are stored in a configuration object rather than being scattered throughout the simulation functions.

source_files <- list.files("R", pattern = "\\.R$", full.names = TRUE)
invisible(lapply(source_files, source))

config <- simgp_tomato_config(seed = 20260512L)

config$population$n_families <- 30L
config$population$lines_per_family <- 12L
config$field$missing_rate <- 0.05

results <- run_simgp_tomato(config, export = TRUE)

See examples/01_custom_parameters.R for a complete worked example.

---

Repository structure

SimGP/
|-- R/                              # Reusable simulation functions
|-- docs/                           # Additional documentation
|-- examples/                       # Example parameter modifications
|-- legacy/                         # Archived monolithic implementation
|-- SimGP_tomato_case_study.R       # Main driver script
|-- README.md
|-- LICENSE
|-- CITATION.cff
|-- NEWS.md
`-- RELEASE_CHECKLIST.md

---

Output files

Running the main script exports the following datasets:

• SimGP_tomato_phenotypes.csv
• SimGP_tomato_snp_matrix.csv
• SimGP_tomato_snp_map.csv
• SimGP_tomato_parental_markers.csv
• SimGP_tomato_pedigree.csv
• SimGP_tomato_true_breeding_values.csv
• SimGP_tomato_causal_qtl_effects.csv
• SimGP_tomato_gxe_effects.csv
• SimGP_tomato_validation_splits.csv
• SimGP_tomato_marker_qc.csv
• SimGP_tomato_sample_qc.csv
• SimGP_tomato_realised_genetic_correlations.csv

A metadata file (README_SimGP_tomato_case_study.txt) is also generated within the output directory.

---

Reproducibility

The default random seed is 20260511.

Running the archived version of SimGP using the same R version and unchanged simulation parameters reproduces the simulated datasets. For formal software releases, it is recommended that the R session information used to generate archived outputs be recorded in docs/sessionInfo.txt.

---

Scope

This release represents the tomato case-study implementation of SimGP rather than a fully general R package or species database.

The framework has been designed so that other plant and animal species can be implemented by modifying species-specific genome structure, breeding system, population design, trait architecture and environmental parameters.

---

Citation

Please cite this software repository if you use SimGP. Where applicable, also cite the SimGP publication.

Software citation

Adebo, U. G. (2026). SimGP: Simulation Framework for Genomic Prediction (Version v1.0.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.21731395

The citation metadata is also provided in CITATION.cff for automatic import into reference managers.

---

Licence

SimGP is released under the MIT License.

See LICENSE for details.

---

Author

Usifo Goodness Adebo

Avecbrio Consult Pty Ltd
Australia