# ToxiCompass external validation data-use notice

## What this website distributes

The website distributes aggregate cohort counts, evaluation metrics,
scientific limitations, and cryptographic provenance hashes for the frozen
ToxiCompass post-hoc robustness snapshot assembled during manuscript revision.
It is not prospective blind validation and does not redistribute row-level
external sequences, source annotations, raw database snapshots, or internal
benchmark overlap identifiers.

This boundary allows readers to inspect the reported evaluation and verify the
identity of the frozen local artifacts without treating ToxiCompass as a mirror
of the source databases.

## Source services

- ConoServer: https://www.conoserver.org/
  - Used for conopeptide external candidates.
  - ConoServer requests citation of the database and restricts its database and
    tools to academic research unless permission is obtained for other use.
- ArachnoServer: http://arachnoserver.org/
  - Used for spider-toxin OOD/challenge candidates.
  - No blanket redistribution permission was identified in the audited local
    source materials, so row-level ArachnoServer-derived records are not hosted.
- Hemolytik 2.0: https://webs.iiitd.edu.in/raghava/hemolytik2/
  - Used for assay-scoped hemolytic/non-hemolytic evidence.
  - Cite Singh et al. (2026), Chemical Research in Toxicology,
    DOI: 10.1021/acs.chemrestox.5c00322.
- IUPHAR/BPS Guide to PHARMACOLOGY:
  https://www.guidetopharmacology.org/
  - Used for therapeutic-peptide and disulfide-rich comparator evidence.
  - The database is provided under ODbL and database contents under
    CC BY-SA 4.0; consult the source website for current terms and citation
    guidance.

Source names and official website graphics are shown only to identify the data
services. Their marks remain the property of their respective owners and do not
imply endorsement of ToxiCompass.

## Scientific interpretation

The comparator labels are evidence-scoped operational roles for this frozen
evaluation. They do not establish universal biological non-toxicity. Stage 2
is a positive-only retained-label ranking analysis; it does not establish
mechanism truth for every sequence or estimate specificity from the positive
cohorts alone.

For additional row-level source data, use the original services and comply with
their current access, citation, and reuse terms.