2 of 9

Open Science Data Repository (OSDR, osdr.nasa.gov)

2 “types” of studies in OSDR:

GeneLab (omics): files are mainly produced by established pipelines, consensus reached via previously-established scientific community standards and AWG community input

ALSDA (non-omics): physiological and phenotypic data does not have established standards, varies widely by assay and even within assays (by equipment, individual researcher, etc). Not much consensus in the scientific community, and not typically expected to be deposited into public repositories

4 of 9

Summary of ALSDA/Phenotypic Data

51 public studies, 17 more in progress

57 public assay metadata configs

40+ additional configs drafted/in progress

Because there is no established metadata for these assays, we draft these ourselves based on published literature and protocols

Drafts are then reviewed by SMEs in our ALSDA AWG community

ALSDA Assays in OSDR

5 of 9

Typical File Types in ALSDA studies

Raw Data Files: images, PDF, log files, csv, etc.

Processed Files: images, PDF, etc.

Data Dictionary Files: csv

Tabular Data Files:

Submitted: results files submitted by the PI, typically in Excel/csv, Prism, or similar statistical program
ALSDA Transformed: A version of the submitted data that has been “tidied”, which is the process that we are trying to refine.

6 of 9

Submitted Data Example

Issues: Multiple tabs, inconsistent ID names, inconsistent spelling and formatting of most things, graphs, formulas, no units specified, text and other characters where integers are expected, and more

7 of 9

ALSDA Transformed Data Example

Same data as previous file, but formatted manually to:

Combine all data into one csv, potentially split by tissue into separate studies
Ensure all column headers are consistent, have a unit, don’t contain acronyms, and more
Contain no metadata columns such as treatment, sex, etc (only results)

This metadata is incorporated into the sample table for each subject

Have a Sample Name that matches the study sample table (often these are new sample names created by OSDR curators) which allows for source tracing across the repo (where applicable, such as Rodent Research animals)
All formulas and graphs removed (retained for reference in the submitted file, which is also attached to the study)
Have non-numerical entries or notes removed, where appropriate (example: NA, ‘not recorded’, etc)

8 of 9

Reasons to Standardize the Transformation Process

Optimize machine readability, for AI/ML and data mining purposes
Optimize reusability (including human readability)
Have unified formatting for future data visualization tools

Current visualization tools, primarily for GeneLab data:

Goal is to develop similar visualization tools for ALSDA study data/assays in the future

9 of 9

List of Questions/Considerations

Plural or singular units? Example: megapascal vs megapascals
Should units be included with the column header (Example: globulin_value_gram_per_deciliter) or included as a separate column?

Capitalization in column headers? All lowercase (globulin_value_gram_per_deciliter), first letter uppercase (Globulin_Value_Gram_Per_Deciliter), other?
General formatting: GlobulinValue_gram_per_deciliter, globulin_value_gram_per_deciliter, other?
Which special characters should we avoid? How should we replace them? Example: β-actin should be beta-actin or ACTB (gene symbol)
What are the contents of your ideal data dictionary? Examples of columns/items to include: acronyms, definitions, ontology link, etc.
For data dictionary files, do you prefer: one large file that applies to all studies and files in the repository, or one file at the study level which only applies to all files in that particular study?

Link to Google Form – please respond to as many questions as possible (9 total, should take less than 10 minutes)

1 of 9