AUTALASSO

AUTALASSO (Automatic Adaptive LASSO) is a software tool for genome-wide prediction (GWP) in animal and plant breeding. It is based on the LASSO (Least Absolute Shrinkage and Selection Operator) technique. It is well-suited for handling sparse problems with many markers and fewer evaluated individuals.

Key features of AUTALASSO:

1. Utilizes the Alternating Direction Method of Multipliers (ADMM) optimization algorithm for efficient computation.
2. Automatically tunes the learning rate using line search and optimizes the regularization factor using Golden section search, eliminating the need for manual hyper-parameter optimization.
3. Provides superior prediction accuracy compared to other methods such as adaptive LASSO, LASSO, and ridge regression implemented in the popular glmnet software, as demonstrated on simulated and real bull data.
4. Offers flexibility and computational efficiency, making it suitable for obtaining high prediction accuracy and genetic gain in GWP.
5. Capable of performing Genome-Wide Association Studies (GWAS) for both additive and dominance effects with smaller prediction errors compared to the ordinary LASSO.