Adaptive First-Order Optimization for Biobank-Scale Genetic Clustering

---

ADAMIXTURE is an unsupervised global ancestry inference method that scales the ADMIXTURE model to biobank-sized datasets. It combines the Expectation–Maximization (EM) framework with the Adam first-order optimizer, enabling parameter updates after a single EM step. This approach accelerates convergence while maintaining comparable or improved accuracy, substantially reducing runtime on large genotype datasets. For more information, we recommend reading our preprint.

The software can be invoked via CLI and has a similar interface to ADMIXTURE (e.g. the output format is completely interchangeable).

System requirements

Hardware requirements

The successful usage of this package requires a computer with enough RAM to be able to handle the large datasets the network has been designed to work with. Due to this, we recommend using compute clusters whenever available to avoid memory issues.

Software requirements

We recommend creating a fresh Python 3.10+ virtual environment. For a faster installation experience, we highly recommend using uv (or pixi). Alternatively, you can use virtualenv or conda.

Important

If you plan to use GPU acceleration, ensure that the CUDA toolkit is correctly loaded (e.g., module load cuda) before starting the installation. This ensures that the dependencies and internal components are correctly configured for your hardware.

As an example, using uv (recommended):

$ uv venv --python 3.10
$ source .venv/bin/activate
$ uv pip install adamixture

Or using virtualenv:

$ virtualenv --python=python3.10 ~/venv/nadmenv
$ source ~/venv/nadmenv/bin/activate
(nadmenv) $ pip install adamixture

Important

macOS Users: ADAMIXTURE requires OpenMP for parallel processing. You must install libomp (e.g., via Homebrew) before installing the package, otherwise the compilation will fail:

$ brew install libomp

Installation Guide

The package can be easily installed in at most a few minutes using pip (make sure to add the --upgrade flag if updating the version):

(nadmenv) $ pip install adamixture

Running ADAMIXTURE

To train a model, simply invoke the following commands from the root directory of the project. For more info about all the arguments, please run adamixture --help. Note that BED, VCF and PGEN are supported:

Tip

GPU Acceleration: Using GPUs greatly speeds up processing and is highly recommended for large datasets. You can specify the hardware to use with the --device parameter:

• For NVIDIA GPUs, use --device gpu (requires CUDA).
• For macOS users with Apple Silicon (M1/M2/M3/M4/M5), use --device mps to enable Metal Performance Shaders (MPS) acceleration.
• Note that biobank-scale datasets are best handled on dedicated CUDA-capable GPUs due to high RAM requirements.

As an example, the following ADMIXTURE call

$ ./admixture snps_data.bed 8 -s 42

would be equivalent in ADAMIXTURE by running

$ adamixture -k 8 --data_path snps_data.bed --save_dir SAVE_PATH --name snps_data -s 42

Two files will be output to the SAVE_PATH directory (the name parameter will be used to create the full filenames):

• A .P file, similar to ADMIXTURE.
• A .Q file, similar to ADMIXTURE.

Logs are printed to the stdout channel by default. If you want to save them to a file, you can use the command tee along with a pipe:

$ adamixture -k 8 ... | tee run.log

Running with multi-threading

To run ADAMIXTURE using multiple CPU threads, use the -t flag:

$ adamixture -k 8 --data_path data.bed --save_dir out/ --name test -t 8

Running with GPU acceleration

To leverage GPU acceleration (highly recommended for large datasets), use the --device flag:

• NVIDIA GPU (CUDA):

  $ adamixture -k 8 --data_path data.bed --save_dir out/ --name test --device gpu

• macOS Apple Silicon (MPS):

  $ adamixture -k 8 --data_path data.bed --save_dir out/ --name test --device mps

Note

Biobank-scale datasets are best handled on dedicated CUDA-capable GPUs.

Tip

Biobank-Scale Execution & High K Values: For large-scale datasets (e.g., UK Biobank, All of Us) with high K values, we recommend the following parameter settings for optimal convergence and performance:

--patience_adam 5 \
--lr_decay 0.85 \
--lr 0.0075

Multi-K Sweep

Instead of running ADAMIXTURE for a single K, you can automatically sweep over a range of K values using --min_k and --max_k. The data is loaded once, and each K is trained sequentially:

$ adamixture --min_k 2 --max_k 10 --data_path snps_data.bed --save_dir SAVE_PATH --name snps_sweep

Other options

• --lr (float, default: 0.005):
  Learning rate used by the Adam optimizer in the EM updates.

• --min_lr (float, default: 1e-6):
  Minimum learning rate used by the Adam optimizer in the EM updates.

• --lr_decay (float, default: 0.5):
  Learning rate decay factor.

• --beta1 (float, default: 0.80):
  Exponential decay rate for the first moment estimates in Adam.

• --beta2 (float, default: 0.88):
  Exponential decay rate for the second moment estimates in Adam.

• --reg_adam (float, default: 1e-8):
  Numerical stability constant (epsilon) for the Adam optimizer.

• --patience_adam (int, default: 2):
  Patience for reducing the learning rate in Adam-EM.

• --tol_adam (float, default: 0.1):
  Tolerance for stopping the Adam-EM algorithm.

• --data_path (str, required):
  Path to the genotype data (BED, VCF or PGEN).

• --save_dir (str, required):
  Directory where the output files will be saved.

• --name (str, required):
  Experiment/model name used as prefix for output files.

• --device (str, default: cpu):
  Target hardware for computation. Choices: cpu, gpu (NVIDIA/CUDA), or mps (Apple Metal).

• -s (int, default: 42):
  Random number generator seed for reproducibility.

• -k (int):
  Number of ancestral populations (clusters) to infer. Required if --min_k/--max_k are not specified.

• --min_k (int):
  Minimum K for a multi-K sweep (inclusive). Must be used together with --max_k.

• --max_k (int):
  Maximum K for a multi-K sweep (inclusive). Must be used together with --min_k.

• --no_freqs (flag):
  If set, the P (allele frequencies) matrix is not saved to disk. Only the Q (admixture proportions) file will be written.

• --max_iter (int, default: 1500):
  Maximum number of Adam-EM iterations.

• --check (int, default: 5):
  Frequency (in iterations) at which the log-likelihood is evaluated.

• --max_als (int, default: 1000):
  Maximum number of iterations for the ALS solver.

• --tol_als (float, default: 1e-4):
  Convergence tolerance for the ALS optimization.

• --power (int, default: 5):
  Number of power iterations used in randomized SVD.

• --tol_svd (float, default: 1e-1):
  Convergence tolerance for the SVD approximation.

• --chunk_size (int, default: 4096):
  Number of SNPs in chunk operations for SVD.

• -t (int, default: 1):
  Number of CPU threads used during execution.

License

This project is licensed under the BSD 3-Clause License - see the LICENSE file for details.

Troubleshooting

CUDA issues

If you get an error similar to the following when using the GPU:

OSError: CUDA_HOME environment variable is not set. Please set it to your CUDA install root.

Simply installing nvcc using conda or mamba should fix it:

$ conda install -c nvidia nvcc

macOS compilation issues

If you get errors related to OpenMP (OMP) during installation on macOS, ensure you have libomp installed via Homebrew:

$ brew install libomp

Cite

When using this software, please cite the following preprint:

@article{saurina2026adamixture,
  title={ADAMIXTURE: Adaptive First-Order Optimization for Biobank-Scale Genetic Clustering},
  author={Saurina-i-Ricos, Joan and Mas Monserrat, Daniel and Ioannidis, Alexander G.},
  journal={bioRxiv},
  year={2026},
  doi={10.64898/2026.02.13.700171},
  url={https://doi.org/10.64898/2026.02.13.700171}
}