Data Release Notes

Versioning Format

The versioning format follows a three-digit structure: X.Y.Z.

• The first digit (X) changes when a major update is introduced, such as changes in the data model.
• The second digit (Y) increments when new data is added.
• The third digit (Z) is updated for bug fixes or minor changes.

Version 3.5.1

Planned: Aug 20th, 2026

Data Updates

• The BMK_000003 dataset (biomarkers_llm_glycan.tsv) has been retired and replaced by a new LLM-mined glycan dataset (BioCompute Object TBA).

Version 3.4.2

Date: Aug 6th, 2026

Biomarker Knowledge Graph

• Fixed an issue where biomarker labels in the KG did not reflect standardized biomarker terminology.
• The JSON-to-NT conversion script now uses controlled vocabulary terms for biomarkers instead of original biomarker names.
• Expanded and corrected edge predicate mappings (biomarker_change, best_biomarker_role, specimen_sampled_from, condition_role_indicator) to use standardized 9-digit OBCI URIs; previously missing predicates have been added, resolving cases where not all biomarker types appeared in "Select Relation."

Biomarker Ontology

• Updated Metadata > Details page: "Biomarker Ontology" now displays as "Ontology for Biomarkers of Clinical Importance (OBCI)."
• Ontology nodes are now sorted alphanumerically.
• Updated hierarchy view instructional text to "Please click on a term on left side to explore more."
• Published a new hierarchy-display-specific ontology version that resolves the multiple-parent issue.

Version 3.4.1

Date: July 23rd, 2026

Data Updates

• Added the caDSR (NCI Cancer Data Standards Registry and Repository) dataset: a pilot set of ~30 manually curated, standardized biomarker entries derived from caDSR permissible values, mapped to the BiomarkerKB schema. Conditions are broadly mapped to general cancer (DOID:162) pending future organ/tissue-specific enrichment. Extraction code and mapping scripts are available on GitHub.

Backend Updates

• Added a new Global Search API (/biomarker/search_global/) enabling search results to be broken out by section (Biomarker Entity, Biomarker Term, Condition/Disease, Specimen, Cross-References, Biomarker ID).
• Each result section now includes an order attribute to control display order on the frontend, standardized as:

 1. Biomarker Entity
 2. Biomarker ID
 3. Biomarker Term
 4. Condition/Disease
 5. Specimen
 6. Cross-References

• Relabeled "crossref" section to "Cross-References" in the UI.

Version 3.3.1

Date: July 13th, 2026

Data Updates

• Split merged datasets back into their original four separate outputs to match the legacy pipeline structure and restore BCO consistency. Affected datasets: BMK_000001, BMK_000005, BMK_000014, BMK_000015.
• Reverted Biomarker ID generation back to the original format (XX1234, two-letter prefix + four-digit numeric ID, e.g., AN6256, AN6256-1), correcting a regression that had introduced a non-standard format (e.g., BMKB151581-1).

Backend Updates

• Automated Docker image cleanup to reduce disk usage on the build server.
• Increased local disk space allocation for the VM.

Bug Fixes

• Fixed column misalignment in old masterlist datasets caused by the missing component_group field. New datasets (e.g., biomarkers_exposure_agents.tsv) include this field; old datasets did not, causing a shift across all rows. Old datasets have been updated to include component_group.

Version 3.2.2

Date: June 22nd, 2026

Data Updates

• The Ontology page has been updated to the latest version.
• The Knowledge Graph files, <all-biomarkers-nt.tar.gz> and <owlnets.tar.gz>, are now up-to-date with the current data.
• Added Dataset Badges to the BiomarkerKB Component Section, linking to data.biomarkerkb.org for all datasets integrated in BiomarkerKB, displayed alongside existing PMID and data source badges.

Backend Updates

• Implemented logic so that search results can be dissected by section (e.g., biomarker component, publication, evidence) and provided this information to the front end for the new intermediate/global search results page. Created a new Global Search API.
• Increased the upload size limit from 2MB to 100MB on the BiomarkerKB Wiki.

Bug Fixes

• All data sources are now displayed correctly under "Data Source" in the Advanced Search.

Version 3.2.1

Date: April 23rd, 2026

Data Updates

• Added a new dataset containing user-submitted biomarkers.
• Added a new manually curated dataset of biomarkers with exposure agents.

Bug Fixes

• Publications and exposure agents are now displayed correctly on the biomarker details pages.

Version 3.1.1

Date: April 9th, 2026

Backend Updates

Complete overhaul of the backend data pipeline architecture:

• Improved ETL processes for greater reliability and scalability
• Enhanced data validation and error handling across pipeline stages
• Optimized performance for faster data processing and reduced runtime
• Refactored codebase for maintainability and extensibility

Version 2.4.3

Date: February 26th, 2026

Data Updates

• Added a new image-based biomarker from the OncoMX dataset.
• Fixed UniProtKB biomarkers that incorrectly included exposure agents.

Backend Updates

• Updated ChEBI API integration to properly parse JSON responses.

Version 2.4.2

Date: February 19th, 2026

Data Updates

• An archive file of all biomarker NTriples is now available for download at data.biomarkerkb.org/BMK_000019.

Version 2.4.1

Date: February 12th, 2026

Data Updates

• A master list of all biomarkers present in BiomarkerKB is now available for download at data.biomarkerkb.org/BMK_000007.

Version 2.4.0

Date: February 5, 2026

Data Updates

• MarkerDB data has been removed due to its license being free for academic use only.

Bug Fixes

• Fixed the issue where glycan biomarkers were being assigned incorrect GlyTouCan IDs in the controlled vocabulary field.
• An advanced search by some data sources, e.g., ClinVar, now yields biomarkers from the data source in question instead of showing all biomarkers.
• Duplicate entity normal range rows have been removed where applicable.
• Entity type casing in searches and search filters has been corrected.
• GWAS and SenNet biomarkers have their controlled vocabulary terms displayed consistently.

Version 2.3.0

Date: January 12, 2026

Data Updates

• New dataset: Top 50 Clinically Relevant Disease Biomarkers created and manually curated by Sparsh Gupta.

Backend Updates

• New Advanced Search type: users can now search biomarkers by Data Source. The following data sources are currently available:
  • cgi (Cancer Genome Interpreter)
  • civic (CIViC)
  • clinvar (ClinVar)
  • edrn (Early Detection Research Network)
  • gwas (Genome-Wide Association Studies)
  • llm_glycan (LLM-extracted glycan biomarkers)
  • markerdb (MarkerDB)
  • mw (Metabolomics Workbench)
  • oncomx (OncoMX)
  • opentargets (OpenTargets)
  • PMC_biomarker_sets (PubMed Central)
  • sennet (SenNet Consortium)
  • top_50 (Top-50 clinically relevant biomarkers)
  • upkb_reviewed_v2 (UniProtKB)

Bug Fixes

• The biomarker_controlled_vocab field in TSV files is now constructed based on the biomarker_id and biomarker_orig tuple. Previously it only used biomarker_id as key, introducing inconsistencies in biomarkers that had multiple biomarker_component objects.

Version 2.2.0

Date: December 22, 2025

Data Updates

• Electronic Health Records data has been added to creatinine biomarkers.
• New dataset: senescence biomarkers from SenNet Consortium.

Backend Updates

• On the API level, each biomarker now contains a new field: biomarker_controlled_vocab which shows the standardized biomarker name. Original biomarker names are now shown in the biomarker_orig field.

Version 2.1.0

Date: December 11, 2025

Data Updates

• Added the LLM-extracted glycan biomarker dataset provided by Cyrus Chun Hong Au Yeung.

Backend Updates

• The incorrect download links on the Data Portal have been fixed.
• LOINC codes are no longer tied to specimen IDs.

Version 2.0.2

Date: December 4, 2025

Bug Fixes

• LOINC codes are no longer tied to specimen (UBERON) IDs.
• For biomarkers that could not be mapped to Controlled Vocabulary the original biomarker name is displayed, followed by "in review".

Version 2.0.1

Data Updates

• Added cross-references to the Common Fund Data Ecosystem (CFDE) Data Coordinating Centers and other resources:
  • GTEx
  • Pharos
  • Reactome
  • Undiagnosed Diseases Network
  • Illuminating the Druggable Genome (IDG) Reactome Portal
  • Metabolomics Workbench
  • SigCom LINCS

Version 2.0.0

Data Updates

• The biomarker field is now standardized using controlled vocabulary terms.
• Added metabolite as an assessed_entity_type to mw_loinc_biomarkers.tsv.
• Added RNAcentral cross-reference support.
• Added Electronic Health Records Normal ranges data from Oracle Health for Troponin I as an example.

Version 1.0.6

Data Updates

• Added a new dataset: MW LOINC biomarkers (mw_loinc_biomarkers.tsv).
• Added National Cancer Institute Thesaurus and Protein Data Bank cross-references.

Backend Updates

• Added the display_name field to the format-converter so data source names appear with correct casing.

Version 1.0.5

Data Updates

• Updated the Troponin biomarker value assessed_biomarker_entity for consistency.
• Added normal ranges from Electronic Health Records provided by the University of New Mexico for Troponin biomarkers.
• Added Cell Ontology and Protein Ontology cross-references.

Backend Updates

• Updated all script paths to use data_source.conf and validated data source names.

Version 1.0.4

This release introduces new datasets, cross-references, and bug fixes.

Data Updates

• Added Cancer Genome Interpreter data on cancer biomarkers from MetaKB.
• Added Metabolomics Workbench LOINC data on metabolite biomarkers.
• Added Cell Ontology and Protein Ontology cross-references.

Bug Fixes

• Fixed issue where cookie preferences weren't being saved when selecting "Allow".

Version 1.0.3

This release introduces new cross-references and updates to ensure compatibility with external resources.

Data Updates

• NCBI cross-references added across gene biomarker entries.
• ChEBI cross-references integrated for small molecules and metabolites.

Backend Updates

• ChEBI API migration: Updated all programmatic links from the legacy SOAP services to the new REST API endpoints, following ChEBI’s platform migration.
  • Old services retired 1 September 2025.
  • New stable API: ChEBI REST API docs
  • New data products and beta interface available at ChEBI 2.0.

Version 1.0.2

Data Updates

• Published updated Metabolomics Workbench data.
• Published sample data from the Early Detection Research Network.

Backend Updates

• evidence_source database names now retain their original casing for accuracy and consistency.
• EDRN identifiers were added to the namespace map.
• HUGO Gene Nomenclature Committee (HGNC) was added to the cross-reference JSON file.
• Fixed an issue where evidence_source values without tags were previously dropped; these are now preserved.
• Added a user-guided spelling correction function to improve data entry quality.
• The TSV-to-JSON converter now automatically checks for header spelling errors.
• Introduced _suggest_header_corrections to flag and propose fixes for misspelled headers.
• Enhanced _stream_tsv with a call to _check_header_spelling to prevent invalid headers from being processed.

Version 1.0.1

Data Updates

• Added xrefs.tsv to the list of datasets.

Backend Updates

• Fixed ID formatting issues in NCBI and UniProt references within oncomx.tsv, removing erroneous spaces (e.g., NCBI: 3288 → NCBI:3288) and extraneous text (e.g., "(composition)"). Affected biomarkers included AN6295-1, AN6756-1, AN6728-1, and others.
• Merged assessed entity type synonyms.

Version 1.0.0

• BiomarkerKB data portal available with OncoMX, OpenTargets, MarkerDB, ClinVar, PubMed Central Biomarker Gene Set Curation, MW, UniProtKB, GWAS, CIViC biomarker data.