Research Data Management

Throughout your research one should be collecting documentation about your experiment and about the contents of your research files/data.  As a result, research documentation can be divided into two types of information – one being experimental information like those contained in laboratory notebooks and two, being metadata or information about the data like README.txt files about your research data files.

Laboratory Notebooks

Laboratory notebooks, which can be either electronic or print, are an essential way and primary record of tracking and accurately recording your research results and process.  Consistency and descriptiveness of your research notes about your research methods, hypotheses, analysis, results, calculations, and statistical methods will increase the integrity and validity of your research especially when one wants to reproduce your experiment or share research at a later date.

• University of Minnesota

  For how to shop for an Electronic Lab Notebook check out the University of Minnesota’s website here for more info on the electronic notebook.

Lab Notebook Features:

Some important information that you might want to capture in your lab notebooks include the following:

• Researcher name
• Project
• Date
• Experiment details --- this includes research methods and purpose
• Name of other data sources used in the experiment

Metadata/README Files

Metadata Definition:

Metadata is information about data; the existence of metadata, whether structured or unstructured, is essential for data to be understood, interpreted, and used. 
 

Metadata Forms:

Metadata documentation may physically manifest itself in the following forms:

• Paper notebook - listing location and names of experimental files
• Digital notebook- embedding hyperlinks to experimental files
• README.txt text file - listing and describing the files contained in the folder on the computer and its abbreviations etc.

• University of Pennsylvania

  University of Pennsylvania has great examples of README.txt files; to view them, look here for more README.txt files from the University of Pennsylvania.

Metadata Breakdown:

Metadata documents data at two levels – research project level and dataset level.

• Research Project Level Metadata sets the context for why the research data was collected and how they were used by researchers.  The following metadata elements help to capture this research context: 
   -  Source: “Data Management General Guidance” - DMPTool
  • Rationale or context for data collection
  • Data collection methods
  • Structure and organization of data files
  • Data sources used
  • Data validation and quality assurance
  • Transformations of data from the raw data through analysis
  • Information on confidentiality, access, and use conditions
• Dataset Level Metadata explain in greater detail the data and the dataset.  The following metadata elements helps to capture this detail:
  -  Source: “Data Management General Guidance” – DMPTool
  • Variable names and descriptions
  • Explanation of codes and classification schemes used
  • Algorithms used to transform the data (may include computer code)
  • File format and software (including version) used

Schemas:

In order to increase the long term discovery, preservation, and understanding of your data at a later date, it is wise to structure your metadata. 

• Metadata structure is often called ‘schema.’  The schema/structure has a defined set of characteristics that describe the dataset.
• Each discipline may have discipline-specific metadata standards or there are generic ones out there. Some discipline specific repositories demand and expect well-structured metadata for those reasons.  It is important that you consult the respective repositories of interest in order to meet their metadata standards.  The UK’s Digital Curation Centre has a list of metadata standards by discipline.

• UK Digital Curation Centre

  Look here to see the UK Digital Curation Centre's metadata standard lists.

• Dublin Core

  One metadata schema that is generic and widely used is the Dublin Core Metadata Element set which contains 15 basic properties.

• DMPTool Template

  Some general aspects of your data can be captured regardless of your discipline or acceptable metadata schema in the following section entitled “How will you document your data?” published by DMPTool.

​Ontologies:

In order to standardize the language to describe your documentation/metadata in order to increase the dataset’s discovery, you might want to use ontologies/controlled vocabularies to describe your data.  Ontologies are shared vocabularies that describe certain aspects/relationships with a respective discipline.  Through using ontologies/controlled vocabularies, you are increasing the user comprehension of your dataset.  The following list, though not exhaustive, are some popular biomedical ontologies:

• Gene Ontology

• Medical Subject Headings (MeSH)

• Chemical Entities of Biological Interest (ChEBI)

Metadata Tools:

• Stanford University lists some handy metadata creation tools, look here to see Stanford's metadata creation tools