A novel convex optimization framework for controlled mass pollination

This repository supports the article “A novel convex optimization framework for controlled mass pollination: capturing additive and non-additive genetic gain in seed orchards.”

The repository is organized into two main workflows:

• 01_CMP_optimization/: user-facing controlled mass pollination (CMP) optimization tools in R and Excel.
• 02_manuscript_reproduction/: manuscript-scale simulation and optimization workflow used to reproduce the study scenarios.

---

Repository structure

CMP-optimization-main/
├── README.md
├── LICENSE
├── 01_CMP_optimization/
│   ├── 001_R/
│   │   ├── CMP_optimization.R
│   │   └── input.xlsx
│   └── 002_Excel/
│       └── CMP_solver.xlsx
└── 02_manuscript_reproduction/
    └── Simulation_optimization_script

---

01_CMP_optimization

This folder contains the main CMP optimization tools for users.

R implementation

Main script:

01_CMP_optimization/001_R/CMP_optimization.R

Example input workbook:

01_CMP_optimization/001_R/input.xlsx

Run the script from the repository root with the bundled input workbook:

Rscript 01_CMP_optimization/001_R/CMP_optimization.R 01_CMP_optimization/001_R/input.xlsx 5 1000

Command-line arguments:

1. Input Excel workbook path
2. Target status number, Ns
3. Total number of operational crosses

The script can also be run without command-line arguments. In that case, it looks for input.xlsx in the current working directory and uses the default settings Ns = 5 and number of crosses = 1000:

cd 01_CMP_optimization/001_R
Rscript CMP_optimization.R

Expected input workbook structure:

Sheet	Required content
Sheet 1	GCA vector, one column, n parents × 1
Sheet 2	SCA matrix, n parents × n parents
Sheet 3	Genomic relationship matrix, n parents × n parents
Sheet 4	Optional cross-limit matrix; 0 = banned cross, values between 0 and 1 = upper bound, blank = no custom limit

The script writes three output sheets back into the same workbook:

Output sheet	Description
Output_Parents_p	Optimized total parental contribution for each parent
Output_Families_Y	Optimized family proportions and operational number of crosses
Operational_Plan_Simple	Editable operational plan for female and male contributions

Excel implementation

Spreadsheet solver file:

01_CMP_optimization/002_Excel/CMP_solver.xlsx

Use this option for a spreadsheet-based CMP optimization workflow. Open CMP_solver.xlsx and follow the instructions inside the workbook to enter inputs and run the solver.

---

02_manuscript_reproduction

This folder contains the manuscript-scale simulation and scenario reproduction workflow.

Main script:

02_manuscript_reproduction/Simulation_optimization_script

Run from the manuscript reproduction folder:

cd 02_manuscript_reproduction
Rscript Simulation_optimization_script

This workflow generates scenario input data, including additive relationship matrices, true GCA/SCA values, half-sib GCA/SCA estimates, and full-sib GCA/SCA estimates. The same script performs CMP optimization for the study scenarios below.

Scenario	Decision information used in optimization
sc1_true	True GCA + true SCA
sc2_hs	Half-sib GCA only
sc3_30	Full-sib GCA + SCA estimated from 30 progeny per cross
sc4_100	Full-sib GCA + SCA estimated from 100 progeny per cross

This script is computationally intensive. For testing, reduce simulation-scale parameters such as n_iterations, nf, snp, nr, reps_vec, and itr_grid before running the full manuscript-scale workflow.

---

Software requirements

R implementation for CMP optimization

Install CRAN packages:

install.packages(c("Matrix", "readxl", "openxlsx"))

The script also requires the gurobi R package, which is installed with Gurobi Optimizer rather than from CRAN. A valid Gurobi installation and license are required.

Excel implementation

The Excel workbook requires Microsoft Excel with Solver support enabled.

Manuscript reproduction workflow

Install CRAN packages:

install.packages(c(
  "dplyr",
  "tidyr",
  "Matrix",
  "reshape2",
  "openxlsx",
  "slam",
  "ggplot2"
))

Additional dependencies require separate installation or licensing:

• asreml: ASReml-R, proprietary software.
• gurobi: Gurobi Optimizer and R package, license required.
• MoBPS, miraculix, and RandomFieldsUtils: install according to the MoBPS package guidance.

---

Contact

Corresponding author: Prof. Milan Lstibůrek
E-mail: lstiburek@fld.czu.cz

First author: Christi Sagariya
E-mail: csagariya@gmail.com