OVO Plugin Development

Please use the Discussions tab on GitHub for questions and support regarding plugin development.

Example plugins:

Tutorials

Developing descriptor plugins for OVO

Part 1: Developing a simple UI plugin for OVO

This section describes how to develop a simple UI plugin for OVO that displays a custom tab on the “🐣 Designs” page in the OVO web app. The limitation of UI plugins is that they cannot introduce additional dependencies, and they have to run “on the fly” in the OVO web app, which means that they cannot run any heavy computations or use any large models.

Starting at 13:48 in the video tutorial above.

Show step-by-step instructions Start by installing latest version of OVO (pip recommended): 
pip install --upgrade ovo

Create your plugin directory using ovo init plugin:

ovo init plugin
# enter a name for your new plugin directory, 
# select "ui" plugin template, 
# and follow the instructions

Install your plugin in editable mode:

pip install -e ./ovo_your_plugin_directory

Now you can start developing your plugin by editing the design_view_....py file in your plugin directory.

This page will be accessible as a tab on the “🐣 Designs” page in the OVO web app.

Please refer to the video tutorial above starting at 13:48 for a step-by-step walkthrough.

See the built-in OVO design pages for inspiration.

Please refer to Streamlit documentation for more information on how to develop Streamlit apps.

 

Part 2: Developing a protein analysis “descriptor” plugin for OVO

This section describes how to develop a protein analysis “descriptor” plugin for OVO, which adds a new tab to the OVO “🐣 Designs” page that allows users to analyze their designs using a custom method.

Starting at 25:09 in the video tutorial above.

Show step-by-step instructions Start by installing latest version of OVO (pip recommended): 
pip install --upgrade ovo

Create your plugin directory using ovo init plugin:

ovo init plugin
# enter a name for your new plugin directory, 
# select "descriptor" plugin template, 
# and follow the instructions

Install your plugin in editable mode:

pip install -e ./ovo_your_plugin_directory

This will add a new tab on the “🐣 Designs” page in the OVO web app where you can submit your workflow and visualize the results.

The next steps will depend on the specific analysis method you want to implement, but in general you will need to:

  • Edit the envs/yourtool.yml file to add any dependencies needed for your analysis method

    • Contribute a Dockerfile to the ovo-containers repository so that we can prepare a pre-built container and make it available for all users to download

  • Edit the shell script in the Nextflow pipelines/.../main.nf file to call your analysis method.

  • Test the workflow using the shell script found in pipelines/.../local/nextflow_test.sh. The workflow should produce a CSV file with descriptor values.

  • For each column in the CSV file that should be saved in the database, create a Descriptor object in the descriptors_yourplugin.py file so that they can be stored in the database and visualized in the UI

  • Now you can submit from the OVO web app and visualize the results

    • If the workflow fails on missing Design IDs, edit the design_id_mapping in the process_results function in models_yourplugin.py to make sure that the ID column in the CSV file from the workflow results is correctly mapped to the Design IDs in the OVO database

To add additional parameters to the workflow:

  • Add the parameters to the pipelines/.../nextflow.config file and the pipelines/.../nextflow_schema.json file

  • Edit the pipelines/.../main.nf file to accept the new parameters and use them in your analysis method

  • Add the new parameters to the test script pipelines/.../local/nextflow_test.sh and make sure the test workflow runs successfully

  • Edit the models_yourplugin.py file to add the parameters to the ParamsType class and set any default values

  • Edit the design_view_yourplugin.py file to add new input fields to the submission dialog window and pass the values to the workflow

Please refer to the video tutorial above starting at 25:09 for a step-by-step walkthrough.

See the built-in OVO design pages for inspiration.

Please refer to Nextflow documentation for instructions how to build Nextflow workflows and the Streamlit documentation for more information on how to develop Streamlit apps.

 

Part 3: Developing a generative “design” plugin for OVO

OVO distinguishes between two types of workflows, “descriptor” workflows that analyze existing designs and “design” workflows that generate new protein sequences and structures.

Developing a “design” plugin is similar to developing a “descriptor” plugin, but it is more complex because it requires implementing (typically) multi-step and multi-dependency Nextflow workflows, and processing logic to create new Design objects in the OVO database based on the workflow results, and a more advanced UI for submission.

A video tutorial and ovo init plugin template for developing a “design” plugin is coming soon. Please use the Discussions tab for questions and support regarding plugin development.