LAGM: Look-Ahead Genomic Mating

This repository contains the code accompanying the unpublished manuscript: "Look-ahead genomic mating optimizes the trade-off between genetic gain and inbreeding in family-based aquaculture breeding programs."

📁 Project Structure

To reproduce the results reported in the study:

1. Directory setup

   Place all scripts in the same directory.

2. Main script execution

   Run the R simulation scripts:

   Rscript Simulation_LAGM.r
   Rscript Simulation_LAGM_fixed.r

   These scripts generate all outputs used in the paper. The raw genomic data must be downloaded manually from NCBI. The preprocessing pipeline that converts the raw sequence data into the format consumed by the simulation lives in the MNNDR repository.

3. LAGM implementation

   The LAGM method is implemented in the lagmRcpp subdirectory. Before running the simulations, install the package from GitHub:

   remotes::install_github("kzy599/LAGM", subdir = "lagmRcpp")

   The default diversity level remains "pair" with unchanged behavior. For diversity_level = "pop", the default is now joint selection, contribution and pairing optimization (pop_epsilon = 0.005), retaining the original gain/diversity normalization searches. Set pop_two_stage = TRUE to restore the legacy J-only search followed by Stage B (mate_allocation_pct); epsilon zero alone does not restore it. See the package documentation for examples, the bounded reward interpretation and reproducible validation.

4. Dependencies

   Before running any script, ensure that all required R libraries are available — they are loaded centrally via utils.r — and that HiBLUP (used for EBV calculation) is installed and configured correctly.

Contact

If you encounter any issues or have questions about the code, please contact: kangziyi1998@163.com