loom2nanopub

Convert TIB Knowledge Loom records into FORRT nanopublications — semantic, signed, replicable assertions published to the nanopub network.

Each Knowledge Loom record becomes three types of nanopublications:

• FORRT Claims — one per statement (what was found)
• FORRT-KL Replication Study — the study design (software, methods, data, script)
• FORRT-KL Replication Outcomes — the results (actual values, metrics, visualizations)

Record mapping

loom_records.txt maps all 32 Knowledge Loom records to their GitLab repos:

a03gjbxh5v  4  lezhnina-2025-1-1   Analysis of difference for Iris species
vjea9aobg7  2  paredes-2022-1      Determinants of B. oleae abundance  DONE
...

Usage

Jupyter notebooks

Single record — notebooks/loom2nanopub.ipynb

Interactive workflow with preview before each publish step:

1. Set LOOM_RESOURCE_ID in the Configuration cell
2. Run all cells top to bottom
3. Preview each TriG file before publishing
4. Mark the record as DONE in loom_records.txt

Batch — notebooks/loom2nanopub_batch.ipynb

Process all non-DONE records:

1. Run Setup cells (section 1)
2. Run Generate — creates TriG files for all records into outputs/<ID>/
3. Review the generated files
4. Set PUBLISH_BATCH = True and run Publish

CLI script

# Single record
python loom2nanopub.py a03gjbxh5v

# All non-DONE records
python loom2nanopub.py --all

# Dry run (show what would be generated)
python loom2nanopub.py --all --dry-run

# Custom output directory
python loom2nanopub.py --all -o ./batch-output/

What gets extracted

The tools parse dtreg JSON-LD files from each record's GitLab repo and the Knowledge Loom API to produce nanopubs with:

Field	Source	Nanopub type
Statement text	KL API	Claim
Source publication DOI	KL API (basises)	Claim
Data source description	dtreg has_input	Claim
Software packages + versions	dtreg executes → part_of	Study
Runtime environment	dtreg part_of → part_of	Study
Method calls	dtreg executes → is_implemented_by	Study
Input data descriptions	dtreg has_input → label	Study, Outcome
Input dataset URLs	metadata.json file_mapping	Study
Analysis script URL	dtreg top-level is_implemented_by	Study
Result values (actual numbers)	dtreg has_output → source_table	Outcome
Target variable	dtreg targets → label	Outcome
Result visualization URL	dtreg has_output → has_expression	Outcome
KL record link	KL URL	Study, Outcome

Publish order

Publishing must be sequential per record because each nanopub references the previous:

1. Claims — published first, resource URIs collected
2. Study — references claims via targetsClaim, study URI collected
3. Outcomes — reference study via isOutcomeOf

Both the batch notebook and CLI handle this automatically.

Requirements

• Python 3.10+
• rdflib
• nanopub (Python library for signing and publishing)
• A nanopub profile (~/.nanopub/ or custom path)

mamba activate sciencelive  # or your env with rdflib + nanopub

FORRT-KL Templates

• Claim: RAu5uTahAxc0OLBB3vaGwK3OQDDZV7QuWtDlBk0Ea3bco
• Study: RALIq4JelUP-q9BuWONcKMJ87B5n59ppcwhQjl-1dheO4
• Outcome: RAw3XdUhxQJfKBaU-cQhV6c7au4rLd5CSUdbMKTS_FB8g

Links

• TIB Knowledge Loom
• Knowledge Loom GitLab
• dtreg paper
• Science Live Platform
• Nanopub network