Agaveis/phenotype-catalog-pipeline: v1.0.0 — PhenotypeCatalog dataset & methodology paper
Year: 2026
Type: software
Abstract: First tagged release of the phenotype-catalog-pipeline reproducibility code, accompanying the PhenotypeCatalog v2 dataset. Artifacts methodology-v1.pdf (8 pages) — methodology paper covering construction, schema, coverage statistics, limitations, intended uses, and reproducibility steps. Released alongside this code, archived to Zenodo via the GitHub integration. scripts/ — six Node.js pipeline stages: Wikipedia people scrape, image URL discovery, Bedrock vision analysis, per-group aggregation, HuggingFace export, phenotype-profile synthesis. prisma/schema.prisma — schema excerpt for the three relevant tables (ethniclist, ethnic_examples, ethnic_image_analysis). prompts/phenotype-analysis-prompt.md — exact vision prompt with design notes. Dataset HuggingFace: https://huggingface.co/datasets/EthnicErotic/phenotype-catalog (CC BY 4.0) 484 ethnic groups · 13,094 notable people · 5,668 per-image phenotype observations Vision model: Anthropic Claude Sonnet 4.6 via AWS Bedrock Total compute cost: $44.66 across 5,668 image analyses Citation Once Zenodo finishes archiving this release (~1 minute), a DOI will be minted and the badge will appear in the README. For now, cite the dataset directly: ```bibtex @misc{phenotype_catalog_pipeline_v1, title = {phenotype-catalog-pipeline v1.0.0}, author = {Jacoby, Jason}, year = {2026}, publisher = {GitHub}, url = {https://github.com/Agaveis/phenotype-catalog-pipeline/releases/tag/v1.0.0} } ``` License Code: Apache 2.0 · Dataset: CC BY 4.0 · Image URLs reference Wikipedia/Wikimedia content under per-image licenses.
Author: Jacoby, Jason
Language: English
Subfield: Statistics, Probability and Uncertainty
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7160488884