Singularity Containers Improve Reproducibility and Ease of Use in Computational Image Analysis Workflows

Abstract

Reproducing computational workflows in image analysis and microscopy can be a daunting task due to different software versions and dependencies. This is especially true for users with little specific knowledge of scientific computation. To overcome these challenges, we introduce Singularity containers as a useful tool to run and share image analysis workflows among many users, even years later after establishing them. Unfortunately, containers are rarely used so far in the image analysis field. To address this lack of use, we provide a detailed step-by-step protocol to package a state-of-the-art segmentation algorithm into a container on a local Windows machine to run the container on a high-performance cluster computer.