borealis-biomodeling / aurora-pkpd

Open Python Web App for Pharmcological Modeling and Analysis

Aurora PK/PD

Open Web App for Pharmcological Modeling and Analysis

Aurora PK/PD is an open-source Python web app implemented using the streamlit framework. Its purpose is to provide a robust and accessible set of pharmacokinetics and pharmacodynamics modeling and analysis tools for academic researchers and small biotech/pharma startups.

You can access the app at

aurora-pkpd.streamlit.app

---

License

This project is licensed under the BSD 2-Clause License - see the LICENSE file for details

---

Contact

• Issues 🐛 : Please open a GitHub Issue to report any problems/bugs with the code or its execution, or to make any feature requests.

• Discussions ❔ : If you have questions, suggestions, or want to discuss anything else related to the project, feel free to use the Discussions board.

---

Supporting

I'm very happy that you've chosen to use Aurora PKPD. This is a project that I'm developing and maintaining on my own time and without external grant funding. If you've found it helpful, here are a few ways you can support its ongoing development:

• Star ⭐ : Show your support by starring the Aurora PK/PD GitHub repository. It helps increase the project's visibility and lets others know it's useful. It also benefits my motivation to continue improving the app!
• Share 📣 : Sharing Aurora PK/PD on your social media, forums, or with your network is another great way to support the project. It helps more people discover the app, which in turn motivates me to keep developing!
• Cite 📚 : Citing or mentioning this software in your work, publications, or projects is another valuable way to support it. It helps spread the word and acknowledges the effort put into its development, which is greatly appreciated!
• Sponsor 💵 : Even small financial contributions, such as spotting me the cost of a tea through Ko-fi so I can get my caffeine fix, can make a big difference! Every little bit can help me continue developing this and other open-source projects. For more details on sponsoring this project see SPONSORING.

---

aurora-pkpd's Issues

Couple GA optimization with Nelder-Mead for curve fitting

It might be interesting to try coupling the GA-based optimization with a Hill climbing method like Nelder-Mead (simplex method) in the optimize.response_fit function. This would work by passing the optimum from the GA to scipy.optimize.minimize for further iteration to the optimum. Since both methods are pretty fast I don't think there should be a significant performance penalty, but this can be assessed.

Build new model button on compartmental upload page not working

When clicking the "Build new model" button on pages/compartmental/upload.py it throws an error:

StreamlitAPIException: Could not find page: build.py.

I haven't checked the code yet, but I'm guessing I forgot to update the path to the build.py file; it should be pages/compartmental/build.py

HT Cell Proliferation page

Complete a basic implementation of the HT Cell Proliferation page; pages/pdanalysis/htproliferation.py.

This page will implement some base functionality using the Thunor Core library to load and analyze high-throughput cell proliferation data. We can link to the Thunor Web application for users needing more advanced features and database support.

Support for molar concentration in the compartmental build page

Currently, the build page pages/compartmental/build.py only supports mass/volume concentration units. I would probably be good to make molar concentrations available too (for more mechanistic models).

New page for drug combination and synergy analysis

Develop and add a new page for analyzing drug combinations and synergy. This could include different metrics of drug synergy, such as combination index, and possibly analysis for the MuSyC framework (https://www.nature.com/articles/s41467-021-24789-z).

Cleanup and better organize the session_state variables

The session_state stores things like the compartmental model and data imported for PD analysis. I think it would be good to organize these better by introducing section-specific dictionaries such as: st.session_state.compartmental that could store variables for that specific section (in this case, the compartmental PK/PD modeling section.). This would help segment the session_state variables and keep things from getting mixed up across the different app pages/sections.

New Roadmap page

Develop and add a new Roadmap page to the app, e.g., as pages/development/roadmap.py
This page can display a roadmap of significant development milestones and their corresponding Issues or Discussions.

This could include creating a new category of pages, "Development," that could also link to this repo and maybe a contributing page.

Compartmental PK/PD Analyze page

Complete an initial implementation of the compartmental PK/PD modeling Analyze page: pages/compartmenal/analyze.py.

This page will implement analyses like local sensitivity analysis and other dynamical model assessments (such as stability analysis, as in https://github.com/blakeaw/MYGO/blob/main/stability-analysis.ipynb).

Linear clearance as an alternative to elimination on build page

Describe the solution you'd like
Option to choose linear clearance with the clearance rate as the input parameter as an alternative to linear elimination in Step 5 of the Compartmental PK/PD Build page.

Include the measurement error/deviation in response fitting

Currently, the optimize.response_fit function does not use the measurement (y-value) error or standard deviation in the curve fitting. However, it would probably be good to allow this by using something like Chi-squared metric instead of SSE when the error values are passed in to the function. The function has the argument to accept the error values but does not actually use it now.

Error in loading model file from GitHub with branch name other than main

Describe the bug
The upload page fails to upload from GitHub when the branch is named master:

To Reproduce
Steps to reproduce the behavior:

1. Go to https://aurora-pkpd.streamlit.app/upload
2. Got to the 'OR, Load from public repository:' section
3. Leave the host set to GitHub
4. Set the User or Organization input to pysb
5. Set the Repo name option to pysb
6. Set the path to pysb/examples/earm_1_0.py
7. Click the Load button.
8. See error: SyntaxError: illegal target for annotation

Expected behavior
The earm_1_0 example model from pysb should be loaded into the app.

Screenshots
If applicable, add screenshots to help explain your problem.

Compartmental PK/PD Fit/Train page

Complete implementation of the compartmental PK/PD modeling "Fit/Train" page: pages/compartmental/fit.py

This page will allow users to upload concentration-time profiles, train PK/PD models to fit the data (model calibration), and estimate relevant PK parameters (e.g., distribution and elimination rate constants).

NCA Page

Complete an implementation of the Non-Compartmental Analysis (NCA) page: pages/noncompartmental/nca.py

Note this first requires the development of the pyNCA package: https://github.com/Borealis-BioModeling/pyNCA