ADNIMERGE2-Package-Building

Introduction

The document describes the procedure to build an R data package from source .csv files with similar workflow of ADNIMERGE2 package.

Building an R Package from csv files

Create R Package Project

Clone the https://github.com/atri-biostats/ADNIMERGE2 repository. This will create the following directories:

data-raw: to store raw-data
R: to store package-specific defined function that includes utils function
man: to store roxygen2 documentation of functions/data in the package
vignettes: to store guidance documents/articles about the package if necessary
tests: to store package related test and retest scripts and will be checked during package building
inst: to store any files that will not be exported to the package, and that will be stored as system file including any external data that is not in R data format (i.e., .rda/.Rdata)
tools: to store auxiliary files that needed during package configuration

Or create an R package project locally, which can be created using usethis::create_package function, and add package-specific metadata into the local package directory, please refer to https://r-pkgs.org/description.html for more information. Once the package-specific metadata is created , copy the pre-defined scripts/functions as necessary from the ADNIMERGE2 github repository with the same file path to your local package project directory.

Download the ADNI Study Data

Download the ADNI study data from the data-shared platform at https://adni.loni.usc.edu/data-samples/adni-data/ either in *.zip or *.csv file format, and store the files in ./data-raw directory

Required to download a data dictionary *.csv file
It is recommended to download files from the data-sharing platform on the same date. If the data is download from Analysis Ready Cohort (ARC) Build, it is recommended to download the file without adding any prefix to the file name as highlighted in the following figure.

Build Package

The next step is start building the package by sourcing ./tools/build.R script file. The ./tools/build.R` script file is used to prepare dataset, generate documentations and finalize package building. The main procedures in this script are presented as follows:

Data preparation:
- ./data-raw/data-prep.R:
  - To store all dataset in ./data directory using usethis::use_data()
  - Some additional data preparation, please see here for more information.
  - Required to specify two input arguments: the data downloaded date (DATA_DOWNLOADED_DATE) in YYYY-MM-DD format and indicator of whether to update any existing data dictionary file (UPDATE_DATADIC)
- ./data-raw/data-prep-recode.R:
  - To map numerically coded values of a dataset based on existing data dictionary as necessary
  - Required to specify indicator argument regarding whether to use the updated data dictionary (USE_UPDATED_DATADIC) if the data dictionary is updated in previous step
- ./tools/generate-derived-data.R:
  - To generate some derived dataset based on the files in the vignettes directory
  - Required to specify the name of derived dataset (DERIVED_DATASET_LIST) as an input argument
Generate data-related documentations:
- To create roxygen2 documentation for the available datasets in the ./data directory.
- ./tools/document.R:
  - To generate documentations for both the raw data and the derived data based on the actual data values, and data dictionary
  - Required two input arguments: indicator of whether to use the updated data dictionary (USE_UPDATED_DATADIC) if it is existed, and name of derived dataset list (DERIVED_DATASET_LIST).
Build the data package: using devtools R package

References

“Alzheimer’s Disease Neuroimaging Initiative — Adni.loni.usc.edu.” https://adni.loni.usc.edu/.