Decoding the Genotype-Phenotype
Map of Vision

An open-source ecosystem curating visual physiology data and applying machine learning approaches to predict opsin phenotypes directly from amino-acid sequences.

Explore Our Tools

Tools & Data Resources

Open-source datasets, interfaces, and computational tools developed by the Visual Physiology DB organization.

VPOD v1.3

The Visual Physiology Opsin Database is a newly compiled database containing 1,714 unique opsin genotypes and corresponding λmax phenotypes collected across all animals from 120+ publications.

View Database Details

VPOD Explorer Interactive

An interactive web version of VPOD for searching records, applying filters, exporting queried data, and suggesting new entries for curator review.

Use the GitHub database for versioned files, notebooks, and reproducible local workflows; use the Explorer when you want to query and download data directly in the browser.

Open Explorer

OPTICS v1.3

The Opsin Phenotype Tool for Inference of Color Sensitivity uses machine learning models trained on VPOD to predict λmax from unaligned sequences and map feature importance to 3D structures.

View Tool

Mission & Roadmap

VisualPhysiologyDB is building an open-source ecosystem for connecting light-sensitive genes, physiological phenotypes, ecological context, and predictive models of light sensitivity.

Long-term Vision

Our goal is to make visual and light-interaction phenotypes accessible, computable, and reusable across scales: from individual amino-acid substitutions to ecological communities and environmental sequence datasets.

The project began with VPOD, which organizes opsin genotype-phenotype data for spectral sensitivity, and OPTICS, which uses VPOD-trained models to predict opsin spectral sensitivity from sequence. Future development will broaden this ecosystem into a more general visual physiology resource for sensory ecology, molecular evolution, biodiversity science, and AI-enabled genotype-phenotype prediction.

Guiding Focus

  • Open licensing, transparent provenance, and reusable datasets.
  • Reproducible workflows that connect gene discovery, phenotype prediction, and ecological analysis.
  • Contributor pathways for data submission, issue reporting, feature requests, and community-developed workflows.

Near-term Priorities

  • 1.Improve VPOD data structure, documentation, validation, and versioning.
  • 2.Improve OPTICS usability, batch prediction, uncertainty reporting, and quality control.
  • 3.Develop OPTICS-Community workflows for predicting light-sensitivity landscapes from species lists, community matrices, and environmental sequence datasets.
  • 4.Provide benchmark workflows, tutorials, example datasets, and standardized outputs for reproducible research.

Longer-term Goals

  • 1.Expand beyond λmax to additional opsin and light-interaction phenotypes, including kinetics, chromophore effects, expression context, and assay metadata.
  • 2.Incorporate ecological and environmental metadata such as habitat, depth, light environment, diel activity, and community composition.
  • 3.Support community-level and ecosystem-level summaries of predicted light sensitivity.
  • 4.Improve interoperability with biodiversity databases, sequence repositories, workflow managers, Galaxy, and downstream modeling tools.

Recent Publications

Accessible and Robust Machine Learning Approaches to Improve the Opsin Genotype-Phenotype Map

Seth A. Frazer & Todd H. Oakley. Molecular Biology & Evolution, 2026.06.23

Read the Publication!

Discovering genotype-phenotype relationships with machine learning and the Visual Physiology Opsin Database (VPOD)

Seth A. Frazer, Mahdi Baghbanzadeh, Ali Rahnavard, Keith A. Crandall, & Todd H Oakley. GigaScience, 2024.09.01.

Read the Publication!