DOME

DOME is a set of community-wide recommendations (ref paper nature methods)  for reporting supervised machine learning-based analyses applied to biological studies.

In the form of questions, the DOME recommendations are divided into four sections: data, optimization, model and evaluation. It’s considered as the minimal requirement for an author to take on consideration while writing a paper to give the paper more reliability and increase transparency and reproducibility. We suggest a list of questions that authors should address in the methods section of manuscripts describing supervised machine learning approaches in order to conform the DOME recommendations and ensure high quality of ML analysis.

The DOME registry facilitates adherence to the DOME Recommendations by transforming them into an accessible online platform. This platform provides guidelines and streamlines the reporting of machine learning methods in biology, thereby enhancing reproducibility and reliability within the broader machine learning community. The DOME registry offers a curated set of annotations for machine learning papers that conform to the DOME Recommendations. Each curated annotation receives a score based on the extent to which the recommendations are followed. Additionally, the DOME registry provides users with the opportunity to contribute annotations using the Data Stewardship Wizard, making collaboration, sharing, and annotation management a simple task.

Bringing the DOME Recommendations to reality through this online platform is a significant step towards enabling critical assessment and improving the reproducibility of published research. The guidelines for curating annotations are fully explained in the DOME Registry.