APRICOT

APRICOT deploys deterministic multicloud virtual infrastructures from Jupyter notebooks to enable reproducible computational experiments in the life sciences.

Key Features:

• Multicloud Infrastructure Deployment: Deterministic setup of virtual infrastructures across multiple cloud environments, supporting elastic clusters that adjust to workload demands, including clusters of PCs.
• Integration with Jupyter Notebooks: Integration with Jupyter notebooks to record infrastructure specifications, data storage, execution, result collection, and infrastructure termination within the notebook.
• Open Science Promotion: Enables deployment of open databases and software tools on existing computational resources to support Open Science practices.
• Reproducibility in Life Sciences: Provides a framework to reduce irreproducibility in preclinical research, a problem reported to affect over 50% of studies.

Scientific Applications:

• Magnetic Resonance Imaging (MRI) Analysis: Processed a real MRI image for prostate cancer characterization by automatically deploying a Message Passing Interface (MPI) cluster.
• Positron Emission Tomography (PET) Image Reconstruction: Performed multiparametric PET image reconstruction with dynamic resource scaling using a batch cluster.

Methodology:

Provides an open-source extension for Jupyter notebooks to specify infrastructure requirements inside computational workflows and automates deployment and management of virtual infrastructures, including automatic deployment of Message Passing Interface (MPI) clusters and dynamic scaling via batch clusters.