SOMAMech somamech

active curating

SOMAMech is a paper-extraction knowledge base for environmental health sciences. It curates publications on PM2.5, smoke and other inhaled exposures and their effects on airway biology into data that validate against the SOMA LinkML schema. One record per publication: each record is a Container that holds the paper’s study subjects, exposures, protocols, assays and measurements, linked to Key Events and Adverse Outcome Pathways. Assay classes cover ciliary function, airway surface liquid, mucociliary clearance, oxidative stress, CFTR ion transport, EGFR signaling, mucin and goblet cell biology, inflammatory markers in BALF and sputum, lung function and gene expression. Every extracted measurement carries an evidence block that quotes the paper, and the quote is checked against the paper text. Papers enter through a stub queue, are claimed one at a time by GitHub issue, and arrive by pull request. SOMAMech follows the DisMech curation pattern.

One record is

a publication, extracted into its assays, measurements and Key Events

40 records as of 2026-10-12

Counted from kb/publications/*.yaml at d8f40d9. Refreshed weekly.

Curation model

Agent-curated	Yes. AI agents generate or maintain most of the content.
Human review	`pull_request`. Agent changes arrive as pull requests that a person reviews and merges.
Agents and tools	Claude Code
Evidence policy	Every extracted measurement carries an evidence block that quotes the paper verbatim. Quotes are checked against the cached paper text. Where only the abstract can be cached, a local check against the full text writes a hash-bound verification receipt, and CI rejects any quote without one.
Validation in CI	schema ontology_terms references
Append-only history	No
Record identifiers	Records are named Container-.yaml and name their source paper by PMID in source_publication. KeyEvent identifiers are shared across papers. Every other entity identifier belongs to exactly one paper.

Features

evidence_snippets Evidence items quote the cited source verbatim.
ontology_grounding Entities bind to public ontology terms with labels.

Grounding and sources

Ontologies and vocabularies: CHEBI GO CL UBERON HP PATO ENVO MONDO NCBITaxon OBI UO PR CLO ECTO ExO XCO

Source	Relation	What is taken
PubMed	prov:hadPrimarySource	The extracted publications. Evidence quotes are checked against cached abstracts and full text.
Zotero library	prov:used	The seed corpus of papers and PDFs, resolved to PMIDs and DOIs in corpus.csv and turned into queue stubs.
SOMA	prov:used	The SOMA data model for Key Event and outcome assays, vendored as the schema.

Products

Product	Kind	Format	Description
Publication extractions `somamech.records`	RecordCorpus	yaml	One Container YAML file per extracted publication under kb/publications/.
SOMA LinkML schema `somamech.schema`	DataModelProduct	linkml linkml
Paper queue `somamech.queue`	OtherProduct	yaml	One stub per candidate paper, with screening status and decision. Stubs are the extraction queue and are not records.

Relations to other Mechs

follows_pattern_of dismech

Contacts

EHS-Data-Standards (Organization) EHS-Data-Standards
Sierra Moxon (Individual) ORCID 0000-0002-8719-7760 smoxon@lbl.gov sierra-moxon

Registry entry

Created 2026-09-22. Last modified 2026-10-01. Edit this entry on GitHub.

SOMAMech is new. The repository was created on 2026-09-17. At commit 2e32037 on 2026-09-22, kb/publications/ held no records. 137 paper stubs sat in the queue and 26 extraction pull requests were open. By commit 56e57cf on 2026-09-29, 8 extractions had been merged, and the maturity moved from seeded to curating on 2026-10-01.

The Mech has no browser site yet. The homepage is the repository. The SOMA documentation at https://ehs-data-standards.github.io/soma/ describes the schema and not this corpus. A pooled Excel workbook is planned as a rolling workbook-latest release and did not exist on 2026-09-22.

The code and records are under the MIT License, as the repository LICENSE states.