Trait BLUP [1,2], a large scale feature selection for genomic prediction using differential evolution; a fairly simple algorithm that assigns fitness based off an individual's GBLUP [3] accuracy.
See main.py
for an example of how to run. This implementation uses numpy matrices throughout.
Convert data accordingly.
See requirements.txt
for dependencies. Use the command pip3 install -r requirements.txt
to install the required libraries.
See generate_sbs.py
to generate SLURM sbatch scripts for various experiments.
[1] Esquivelzeta-Rabell, C & Al-Mamun, Hawlader & Lee, Sang-Han & K Lee, H & D Song, K & Gondro, Cedric. (2015). Evolving to the Best SNP Panel for Hanwoo Breed Proportion Estimates.
[2] Al-Mamun, Hawlader & Kwan, Paul & Clark, Sam & Lee, Sang-Han & Lee, H K & Gondro, Cedric. (2015). Genomic Best Linear Unbiased Prediction Using Differential Evolution.
[3] Clark S.A., van der Werf J. (2013) Genomic Best Linear Unbiased Prediction (gBLUP) for the Estimation of Genomic Breeding Values. In: Gondro C., van der Werf J., Hayes B. (eds) Genome-Wide Association Studies and Genomic Prediction. Methods in Molecular Biology (Methods and Protocols), vol 1019. Humana Press, Totowa, NJ