Here at the Khan Lab, we encourage and support open-science practices.
We strive to release the raw, de-identified data for any study where we are collecting data, or any derived data that we have generated. The BIDS (Brain Imaging Data Standard) is an open standard for describing neuroimaging data. We have tools that help to automatically convert acquired data in to this standard (tar2bids, autobids)
To support reproducibility, we strongly encourage the containerization of workflows so that processing can be reproduced on any system, regardless of software / hardware dependencies. Singularity is a container software designed for scientific computing and used for:
- Ensuring the same software setup can be used on different machines
- Deploying and sharing applications (BIDS Apps) for standardized image processing
We also encourage the use of Snakemake for end-to-end workflow management. This is a tool for scripting your entire workflow, from raw data to figures in your papers. Snakemake allows you to easily scale up your analysis from a single subjec to an entire dataset.
We generally publish pre-prints of all new manuscripts on biorxiv - this is done at the same time as submission to a journal.
These are documents, written, and registered (not visible to the public until released) either before you collect data or carry out analyses, which will describe the analyses that you will do. This is to ensure that one does not hypothesize about the data after the results are known (HARKing), and more clearly delineates hypotheses from post-hoc tests. While this has not yet become common practice in our lab, we are moving towards this!