Skip to content

Latest commit

 

History

History
180 lines (141 loc) · 18.3 KB

paper.md

File metadata and controls

180 lines (141 loc) · 18.3 KB
title tags date output authors bibliography editor_options affiliations
NPSdataverse: a suite of R packages for data processing, authoring Ecological Metadata Language metadata, checking data-metadata congruence, and accessing data
R
metadata
Ecological Metadata Language
data
National Park Service
Environmental Data Initiative
machine readable
Open
FAIR
DataStore
data package
data publication
data access
21 October 2024
word_document html_document pdf_document
default
df_print
paged
default
name orcid equal-contrib affiliation
Robert L. Baker
0000-0001-7591-5035
true
1
name orcid equal-contrib affiliation
Collin Smith
0000-0003-2261-9931
true
2, 3
name orcid equal-contrib affiliation
Sarah E. Wright
0009-0004-5060-2189
true
1
name orcid equal-contrib affiliation
Issac Quevedo
0000-0003-0129-981X
true
4
name orcid equal-contrib affiliation
Kristin Vanderbilt
0000-0003-1439-2204
true
1
name orcid affiliation
Carl Boettiger
0000-0002-1642-628X
5
name orcid equal-contrib affiliation
Judd M. Patterson
0000-0002-0951-7917
true
1
name orcid equal-contrib affiliation
Joe DeVivo
0000-0002-0414-7865
true
1
paper.bib
markdown
wrap
72
name index
National Park Service, USA
1
name index
Environmental Data Initiative, USA
2
name index
University of Wisconsin, USA
3
name index
Student Conservation Association, USA
4
name index
University of California, Berkeley, USA
5

Summary

The NPSdataverse is a suite of R packages developed to create, document, publish, and access data and metadata in open and machine-readable format. NPSdataverse is modeled off of the tidyverse concept of several packages built with a common goal [@Wickham2019]. The NPSdataverse supports Ecological Metadata Language (EML) metadata and .csv data files. Some of the constituent R packages (EML and EMLassemblyline) are general-use and aimed at authoring EML documents. Other R packages (QCkit, EMLeditor, DPchecker and NPSutils) are designed and maintained by the National Park Service (NPS). Although many functions within the NPSdataverse packages are NPS-specific (particularly some API calls), whenever possible the functions are written so that they can also be used by the general public. Scientists conducting permitted research in NPS units can utilize the NPSdataverse to efficiently and consistently meet the data delivery requirements of their permits. Additionally, the packages will be useful for data management plans in a wide variety of grant proposals and for anyone that needs to create open data and machine readable metadata. Finally, the ability to swiftly and easily author, edit, and check Ecological Metadata Language (EML) metadata in a reproducible fashion will be useful for data publication at any number of repositories or data journals.

Statement of Need

Following a movement for transparency in scientific research and data accessibility, the U.S. implemented the federal OPEN Government Data Act [@OpenData2018]. The Open Data Act mandates that federal agencies provide data in open formats with metadata. Subsequently, many funding agencies such as the National Science Foundation have required grant awardees make data public, often including metadata [@nsf2015]. Multiple publishers have followed suit [@Wiley2022; @Springer2023] and require data availability statements upon publication.

One goal of open science, and requirement of the recent "Nelson Memo" from the U.S. Office of Science and Technology Policy [@Nelson2022] is to make data FAIR: findable, inter-operable, accessible, and reuseable [@Wilkinson2016]. These goals are often achieved by including structured, machine-readable metadata that conforms to a defined schema along with the data. Ecological Metadata Language Metadata (EML) is one metadata standard that is particularly amenable to studies with rich taxonomy [@Jones2006; @EML2019]. It has been adopted by multiple research organizations including the Ecological Data Initiative (EDI), National Ecological Observatory Network (NEON), Global Biodiversity Information Facility (GBIF), Swedish Biodiversity Data Infrastructure (SBDI), French Biodiversity Hub ("Pole National de Donnees de Biodiversite"), U.S. National Park Service, and others.

Nevertheless, actual availability of data and metadata varies [@Federer2018; @Tedersoo2021], perhaps because there is a need for more infrastructure and tools to meet the goals of open data and open science [@Huston2019]. Multiple solutions have been presented, including ezEML, a tool for authoring metadata in Ecological Metadata Language and publishing data and metadata to a repository [@Vanderbilt2022]. ezEML has an intuitive graphical user interface with a relatively low learning curve; however, it does have some drawbacks. For instance, ezEML is not scriptable, which makes repeated deployments of the same or similar workflows challenging and can limit reproducibility. ezEML also requires that the user upload their data to an external site for processing, which may not be suitable for sensitive data. Here we introduce the NPSdataverse, a series of R packages for authoring, editing, and checking EML metadata locally in a robust, repeatable, and scriptable fashion. R Packages within the NPSdataverse leverage earlier work using R to create and manipulate XML based EML files [@Boettiger2019]. Building upon that framework, we add user-friendly EML creation workflows; integration with taxonomic databases; fast, easy editing of existing metadata; congruence checks to test correspondence between data and metadata; and integration with public repositories such as the National Park Service's DataStore. The EML metadata file in .xml format along with the .csv data files it describes comprise a "data package". In addition, R packages within the NPSdataverse also include data functions that expedite quality control, facilitate interoperability, provide the ability to download data directly from DataStore, and leverage the rich EML associated with the data regardless of repository of origin.

NPSdataverse R package

The NPSdataverse package is a meta-package that loads packages within the NPSdataverse into R [@Baker_NPSdataverse2024]. It provides a convenient way to download, install, and load many of the R packages needed to create and access data packages consisting of rich Ecological Metadata Language metadata and .csv data files:

pak::pkg_install("nationalparkservice/NPSdataverse")
library(NPSdataverse)

NPSdataverse will automatically check that the latest version of each R package is being loaded: either from the main development branch on GitHub.com or the latest version on CRAN. If updates are indicated, the user will be alerted and given instructions on how to update the relevant packages. To prevent API limits at GitHub (and to facilitate scripted workflows such as those at High Performance Computing facilities), NPSdataverse only checks for updates from an interactive R session and will skip checks when the system is not on-line or GitHub.com is not responding.

QCkit R package

QCkit ("Quality Control kit") is primarily a data processing package designed to prepare data for metadata creation and publication [@Baker_QCkit2024]. This package serves two main functions: 1) Providing a suite of data quality control functions to be used across datasets regardless of the project, and 2) a suite of functions to apply data standards that promotes interoperability among datasets. For instance, QCkit includes functions that can help manage date-time formatting, can check data files for threatened or endangered species, and can help increase inter-operability by suggesting appropriate Darwin Core standards for naming data. QCkit also facilitates documenting data processing with functions that can generate a DataStore reference based on GitHub.com releases. The DataStore reference can hold processing scripts, code, or packages and have Digital Object Identifiers (DOIs) attached to them that are registered with DataCite once the DataStore reference is activated. QCkit is designed as an expandable framework that can adapt to new quality control tests or as new data standards are adopted.

EML R package

The EML ("Ecological Metadata Language") package is a fundamental package that allows for importing .xml files, creating and validating validating EML within R, and writing R objects back out to .xml files [@Boettiger2024]. EML allows for creating fully fledged Ecological Metadata Language Metadata files using nested S3 lists within R while relying on the R/emld package [@Boettiger2019_emld].

EMLassembyline R package

The EMLassemblyline package builds upon EML and adds substantial functionality [@Smith2022]. For instance, EMLassemblyline allows the user to supply .csv files, which are used to generate template .txt files. Users can adjust the template files as needed and use the EMLassemblyline::make_eml() function to generate an R-object that can be exported via EML as an EML-fomatted .xml file. EMLassemblyline includes the ability to generate entire taxonomic backbones from lists of scientific names via API calls to ITIS, GBIF, or Worms. EMLassemblyline will validate the R object against the EML schema and provide helpful hints on what might have gone wrong during the EMLassemblyline::make_eml() process. EMLassemblyline provides an efficient bridge between .csv data and EML metadata for users who are familiar with R but may not be experts on the EML schema or the detailed nested lists needed to create EML within R via the EML package. Products from the EMLassemblyline pipeline are suitable for publication at multiple repositories including the Environmental Data Initiative.

EMLeditor R package

The EMLeditor package allows users to quickly and easily view components of metadata in R and make on-the-fly edits to metadata [@Baker_EMLeditor2024]. Edits made to EML objects using EMLeditor do not require re-running the EMLassemblyline functions to make EML. This is a significant improvement because running EMLassemblyline functions can be time consuming, especially if there are many taxa that need to be resolved. EMLeditor includes the ability to pick specific licenses (CC0, CC-BY, etc), add ORCIDs, include organizations as authors, and much more. EMLeditor also adds specific content necessary to be compliant with NPS's DataStore. With the proper permissions, EMLeditor can be used to generate draft references and reserve DOIs on DataStore as well as upload data and metadata files to DataStore. Finally, EMLeditor contains a .rmd template file that, after loading the package, is accessible in Rstudio under Files > New File > R markdown. The template provides an editable script that walks the user through using EMLassemblyline, EMLeditor, and DPchecker to create and validate EML metadata in R.

EMLeditor "set" class functions (which includes all functions that begin with "set_" such as "EMLeditor::set_abstract()") will add several NPS-specific items to the metadata using their default settings. For instance, these functions will set NPS as the publisher, Fort Collins as the publication location, and will add a "for or by NPS = TRUE" statement to the metadata. To invoke these functions without adding the NPS-specific metadata elements, set the parameter NPS = FALSE when calling each "set_" class function. Non-NPS publisher information can be added using the EMLeditor::set_publisher() function with the parameters for_or_by_NPS and NPS set to FALSE:

#set the abstract without NPS-specific information:

new_metadata1 <- set_abstract(eml_object = old_metadata,
                            abstract = "This is example abstract text",
                            NPS = FALSE)

#add custom publisher information:

new_metadata2 <- set_publisher(eml_object = new_metadata1,
                            org_name = "My Institution",
                            street_address = "1234 Sesame St.",
                            city = "Anytown",
                            State = "Delaware",
                            zip_code = "12345",
                            country = "USA",
                            URL = "https://www.myinstitution.us",
                            email = "[email protected]",
                            ror_id = "",
                            for_or_by_NPS = FALSE,
                            NPS = FALSE)

By default, EMLeditor functions provide verbose user feedback and may require user input to confirm some operations. These checks are intended to help guide users, prevent inadvertent mistakes, and limit unnecessary API calls. However, requiring user input can hamper highly scripted approaches and limits reproducability. Therefore, all EMLeditor functions can be set to circumvent these requirements using the parameter force = FALSE.

#example setting the abstract while suppressing user feedback and input:

new_metadata <- set_abstract(eml_object = old_metadata,
                             abstract = "This is example abstract text",
                             force = TRUE)

DPchecker R Package

The DPchecker ("Data Package checker") package provides feedback on data-metadata congruence [@Baker_DPchecker2024]. Here, a "data package" consists of the EML metadata file with a filename that ends in *_metadata.xml and one or more data files in .csv format, all of which are in a single directory (and the directory contains no extraneous .csv or .xml files). DPchecker is useful for both data package authors and reviewers. DPchecker goes beyond validating EML objects in R against the EML schema. Using the DPchecker::run_congruence_checks function, DPchecker will conduct a series of 46 tests. These are divided into several categories to check whether:

  1. Metadata are well formatted (file names are not duplicated, files specify the field delimiter, data files have URLs, the proper delimiter and header row numbers are present, etc.).
  2. Metadata elements necessary for DataStore automated extraction are present (creators have valid surnames, publication date is present and in the correct ISO-8601 format, keywords are present, abstract and methods are present and well formatted, etc).
  3. Recommended EML elements are present including ORCiDs and a notes section.
  4. Metadata and data are in congruence including all files listed in metadata refer to data files, the columns in the metadata match the columns in the data files, missing fields in data files are properly documented in metadata, and dates in data files fall within the date ranges given in the metadata, etc.
  5. Data and metadata are in compliance with (a subset of) federal regulations including tests for information that should not be released to the public such as non-.gov emails and GPS coordinates for restricted data packages.

For each test, the data package may fail with an error, fail with a warning, or pass. When possible, warnings and error messages indicate the appropriate EMLeditor function to address the problem. DPchecker will often throw a warning even if an EML element exists and is properly formatted but could by improved to increase the FAIR characteristics of the metadata. For instance, DPchecker will throw a warning if an abstract is less than 20 words long as it is unlikely the creator is able to meaningfully describe the data collection and processing in less than 20 words.

NPSutils R Package

The [NPSutils](https://nationalparkservice.github.io/NPSutils/) ("NPS utilities") package serves primarily as a way to access data [@Baker_NPSutils2024]. NPSutils provides avenues for directly downloading data from DataStore using R. NPSutils can also import data downloaded from any repository into R and take advantage of rich EML metadata to call column types. NPSutils provides some basic meta-analysis capability, assuming certain interoperabilty standards are met (such as consistently naming columns with Darwin Core parameters or other domain-accepted parameter names). NPSutils can also be used to import data and metadata into common data visualization tools.

Example of how to download and access data:

# download a data package from datastore:
# the data package will be downloaded to ./data/2300498

NPSutils::get_data_package(2300498)

# load the data package into R:
# returns a list of tibbles; each tibble corresponds to a single data file

mojn <- NPSutils::load_data_package(2300498, assign_attributes = TRUE)

Acknowledgements

We acknowledge contributions from across the National Park Service, but in particular from the Inventory and Monitoring Division. Members of the NPS Long Term Data Management Governing Board provided critical guidance and insight (in addition to several of the authors, these include Kristen Bonebrake, Adam Kozlowski, Ryan Monello, Mark Isley, and Megan Swan). Justin Mills (currently at U.S. Fish and Wildlife Service) and Derrick Dardano helped with navigating API and Active Directory interfaces, Marsha Leavitt made and explained numerous updates to DataStore. Dan Gussett, Kate Miller, and Pete Budde facilitated software availability, and Meg White supported and endorsed the project. We are particularly indebted to our strong user base and their very helpful feedback including Alison Loar, Christina Appleby, Kirk Sherrill, Lisa Nelson and Tom Phillipi. Numerous Student Conservation Association interns made contributions to the code base including Sarah Kelso, James Brown, and Amy Sherman. Alissa Graff (currently at the Internal Revenue Service) provided important input on early versions of NPSutils.

References