In maelstrom-research/harmor: Dataset Harmonization Utils

The harmonization process prerequisites are:

1. Having access to a Opal server with write premission in a project and premission to use R,
2. Having a definition of the targetted data schema,
3. Having access to one or more study-specific dataset that will be used to build the harmonized dataset.

Implement the Data Schema

The targetted data schema must be prepared: definition of the targeted harmonization variables with their proper annotations. This can be done by different ways:

• directly in Opal,
• or in R, then saved in Opal.

Functions that can be used: getOpalTable(), annotate(), annotations(), saveOpalTable().

Generate the Harmonization Books

From the data schema, one book per study will be generated. Each harmonization book will have placeholders for:

• the description of the harmonization scientific decisions (literature references, methods, etc.),
• the R scripts for deriving study-specific data into each of the harmonized variables.

The harmonization book can be designed for one or more domains of interest (Lifetyle and behaviors, Diseases, etc.).

Functions that can be used: makeHarmonizationBook().

Implement the Harmonization Books

In the R markdown files of the harmonization book, implement:

• the extraction of the study-specific datasets which are the data to be derived,
• the derivation of the study-specific variables to the harmonized ones,
• the annotation of each harmonized variable with harmonization status and description,
• the saving of the resulting dataset in a Opal project.

Functions that can be used: getOpalTable(), annotate(), annotations(), saveOpalTable().

Execute the Harmonization Book

The book will execute the R code chuncks that will perform the harmonization for the study, resulting in one or more datasets saved as Opal tables.

maelstrom-research/harmor documentation built on Dec. 6, 2019, 7:31 p.m.