Skip to content

Data Standards and De-identification

This page covers the published standards ImmPort templates are built on, how values are checked against controlled vocabularies, and the de-identification rules for human subject data.

Published standards behind the templates

Standard Covers Source
HIPC Bioinformatics Standards Working Group guidelines Flow Cytometry / Mass Cytometry (CyTOF), Metabolomics & Proteomics Mass Spectrometry, Transcriptional Profiling, Subject Immune Exposure See HIPC guideline-to-template mapping below
NIH-endorsed Common Data Elements (CDEs) Many study, subject and clinical fields Common Data Elements. ImmPort does the mapping when it loads your data, so you don't need to do anything

When a template and a published standard appear to disagree

The template's own field constraints (required/conditional, controlled vocabulary, length) always win for validation purposes — the published standard explains why a field exists, not what will pass the validator.

HIPC Bioinformatics Standards Working Group Guidelines

ImmPort works with the HIPC Bioinformatics Standards Working Group to develop guidelines for metadata and results.

How controlled vocabulary is checked

Any template column marked CONTROLLED VOCABULARY in its requirements (see How to Read the Template Documentation) is checked against an lk_* lookup table at parse time:

  • The check is case-insensitive — "Male", "male", and "MALE" all pass if Male is a valid lk_human_sex entry.
  • On success, the value is rewritten to the lookup table's canonical case spelling — don't assume the byte-for-byte value you submitted is what ends up stored/returned.
  • On failure, validation fails with an error referencing the field and its lookup table.
  • Some columns additionally have a preferred vocabulary — a secondary, non-blocking normalization layer (e.g. population marker names) separate from the required-match controlled vocabulary check.

Where to browse the actual value lists

  • Ontologies overview — a searchable table of every lk_* table, the ontology it's sourced from, and which ImmPort table/column references it.
  • Lookup Tables under Template Documentation — per-table browsable term lists (the same tables the controlled-vocabulary check above validates against).

Don't hardcode a vocabulary snapshot

Lookup tables are maintained independently of template schema versions and can gain new terms between releases. If you cache a vocabulary list locally, re-fetch it periodically (or re-check via Ontologies) rather than assuming a list captured once is exhaustive.

Subject de-identification

ImmPort follows the HHS Safe Harbor method. When you fill in templates (rule R-19):

  • No direct identifiers. Don't enter names, phone or fax numbers, email addresses, account numbers or any other Safe Harbor identifier, including in free-text description and comment fields.
  • No calendar dates for subjects. Give time points and schedules as study days relative to day 0, or to a reference day you define.
  • No location below state level. Location fields must not contain a city, county or zip code.
  • Non-identifying subject IDs. Your user-defined subject IDs must not be traceable to your source system's identifiers.
  • Ages over 89. Enter the true age in years. The uploader changes any age above 89 to 90 and appends ;age rounded down to 90 by ImmPort Upload tool to the subject's description. Don't round ages yourself, or the record loses the note that rounding happened.

In addition, ImmPort:

  • never displays or exports subject, sample or experiment sample source identifiers, which cuts the link to your system;
  • doesn't share original unstructured result files unless the provider agrees to the disconnect from source identifiers, or the data can be parsed into templates with identifiers hidden;
  • reviews non-template clinical files (for example from closed trials) to remove identifying information, shift dates to study day 0, and drop comment fields that might identify subjects. Files from ITN TrialShare are already de-identified.

Human genomic data

ImmPort doesn't normally accept GWAS-level human SNP genotyping or next-generation human genome sequencing data. Deposit it in dbGaP or SRA and link to it with the Public Repositories template.