# Character Similarity Dataset: Phenoscape Ontology Trait Descriptions and Pairwise Semantic Similarity for Fishes and Vertebrates

## Facts

- Provider: Imageomics Institute (Jim Balhoff, Soumyashree Kar, Juan Garcia, Hilmar Lapp)
- License: CC-BY-3.0 (https://huggingface.co/datasets/imageomics/char-sim-data)
- Size: 24 files
- Format: TSV (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/imageomics/char-sim-data
- Access: public, readable without an account
- DOI: https://doi.org/10.57967/hf/6891
- Category: Genomics & Bioinformatics
- Source: https://huggingface.co/datasets/imageomics/char-sim-data
- Page: https://scigantic.com/archive/708c261b-88a9-54ed-8218-8d4aa30f71d1
- Keywords: Genomics & Bioinformatics, TSV, Hugging Face

## About

About 19,000 textual descriptions of anatomical traits (phenotypic characters) of fishes and other vertebrates, taken from the Phenoscape Knowledgebase, paired with ontology-based semantic similarity scores between trait descriptions (maxIC, Jaccard and SimGIC). The Phenoscape Knowledgebase is built on OBO ontologies, chiefly the Uberon anatomy ontology, so the scores measure how close two phenotype annotations are in ontology space. The dataset was built to train and evaluate text encoders that align free-text character descriptions with ontological similarity; the reference model Trait2Vec reaches a Spearman correlation of 0.7057 on the test set. raw-source/ holds the Phenoscape build files phenex-data-merged.ofn.gz (OWL functional syntax) and phenoscape-kb-tbox-classified.ttl.gz (Turtle); processed-data/ holds gzipped TSV pair tables split into training, all-non-train and non-overlap sets (80 percent training) for four fish orders (Characiformes, Cypriniformes, Gymnotiformes, Siluriformes) and their union. About 41 GB. Licensed CC BY 3.0 by the Imageomics Institute; cite the dataset (Balhoff, Kar, Garcia, Lapp, doi:10.57967/hf/6891) and Phenoscape.

## Load the data

```python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="imageomics/char-sim-data", filename="processed-data/all/data_10p_ALL_NON_TRAIN.tsv.gz", repo_type="dataset")
print(path)
```

## Data structure

File types in sampled files: TSV

Example file: `processed-data/all/data_10p_ALL_NON_TRAIN.tsv.gz`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/708c261b-88a9-54ed-8218-8d4aa30f71d1
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/708c261b-88a9-54ed-8218-8d4aa30f71d1
- Over MCP: `get_schema_card` with id `708c261b-88a9-54ed-8218-8d4aa30f71d1` at https://api.scigantic.com/mcp

## Related topics

- [Genomics & Bioinformatics datasets](https://scigantic.com/topics/genomics-and-bioinformatics)
- [TSV datasets](https://scigantic.com/topics/tsv)
