Dictionary Generation and Dataset Versioning

A license is required to access the Data Profiler on the DNAnexus Platform. For more information, please contact DNAnexus Sales (via [e-mail]).

Dictionaries and Coding File Generation

Data Profiler can automatically generate a data dictionary, entity dictionary, and coding file for your dataset. These outputs help standardize dataset metadata and support downstream workflows such as Apollo ingestion through the Data Model Loader (DML).

If you do not provide a data dictionary or coding file as input, Data Profiler will automatically infer them from the dataset during profiling. If these files are provided, the application will use the supplied metadata and coding definitions during profiling.

The generated files include:

• Data Dictionary — Column-level metadata describing table columns, data types, and schema information.

• Entity Dictionary — Table-level metadata describing entities and dataset structure.

• Coding File — Information about coded or categorical values detected across tables.

These files are automatically generated during profiling, helping ensure datasets are standardized, ingestion-ready, and easier to govern across downstream systems and analyses.

You can also edit the generated data dictionary, entity dictionary, and coding file, then re-profile the dataset using the updated metadata. To do this, navigate to Data Profiler → Edit Schema, make the required changes, and click Profile.

Data Profiler allows you to edit schema/dictionary and reprofile the dataset

Reprofiling and Dataset Versioning

When you re-profile a dataset, Data Profiler applies the updated data dictionary, entity dictionary, and coding file to generate a new version of the dataset. The original dataset and its profiling results remain unchanged, preserving full traceability.

Each re-profile operation creates a new dataset version, allowing you to:

• Track how the dataset evolves as metadata is refined.

• Compare profiling results across different versions.

• Preserve the history of schema and coding changes without overwriting previous results.

You can view all versions of a dataset through the Dataset Lineage, which provides a complete history of the profiling and re-profiling process. This enables you to trace the evolution of your dataset and easily navigate between previous and current versions.