-
Notifications
You must be signed in to change notification settings - Fork 32
GSoC 2025
Welcome to Open Free Energy's Google Summer of Code Ideas Page!
If you're interested in working with us:
- Introduce yourself on our discussions page.
- Try out OpenFE
- Check out our Project Ideas
- Submit a project pre-proposal
Because OpenFE is a scientific software package, most of our projects require that a contributor has some degree of domain knowledge (molecular dynamics simulations, free energy calculations, etc.) to quickly get up-to-speed on our code base. To evaluate whether your skills and knowledge base are a good fit for a project before you commit to writing a full proposal, we ask that you submit a project pre-proposal here.
If we reject your pre-proposal, we hope we've saved you the trouble of a full proposal, and that you find another GSoC project more suited to your background. If we accept your pre-proposal, great - you should continue your application to work with us!
Many of the projects below require some knowledge of free energy calculation workflows and adjacent concepts, so please take note of the "desired skills" section of the project(s).
These projects are problems scoped out by our team, but if you have your own proposal, please get in touch!
Time Estimate: 350 hrs
OpenFE and our users are always looking to introduce new protocols, but testing and validation of these protocols can be a barrier to adopting new methods. We are looking to build tooling to make validation and testing of protocols a standardized process.
A new testing package that makes it simpler to execute a standardized validation suite for free energy protocols.
- MD simulation knowledge
- Regression and Integration testing
- Data management and storage best practices
@atravitz, @ethanholz, @ialibay
Time Estimate: 350 hrs
Benchmarking is critical for evaluating OpenFE's accuracy and computational efficiency, and having benchmarking results be public and reproducible is essential. Currently, benchmarking is performed according to a detailed set of instructions, but we are looking to automate much of this process to improve reproducibility and minimize human error.
We expect this new tool might function similarly to open force field's https://github.com/openforcefield/openff-qcsubmit, and could take advantage of alchemiscale for execution.
A new tool that automates the benchmarking process, including results upload.
- MD simulation knowledge
- Regression and Integration testing
@atravitz, @ethanholz, @ialibay
Time Estimate: 350 hrs
Analyzing the results of binding free energy calculations can give insights into what drives the potency or selectivity of ligands. Interaction fingerprints encode the interactions observed in 3D structural data in a vector representation. These fingerprints can then be used to investigate and compare protein-ligand interactions in a drug discvery campaign. This project aims to add such tooling for the analysis of binding free energy calculations using the opensource tool ProLIF.
New feature in openfe_analysis that interfaces with prolif
- MD simulation knowledge
- MD trajectory analysis experience
@hannahbaumann, @alyssatravitz
Time Estimate: 175 hrs
Cinnabar is the package OpenFE uses to report free energy calculation results. We are looking to add novel metrics for ranking and classification, such as top N% ranking, and fraction of best ligands.
- New functionality in cinnabar
- Improved cinnabar user experience
- Free energy calculation experience
- MD simulation analysis
@hannahbaumann, @joshhorton, @atravitz
Time Estimate: 350 hrs
Having sufficient overlap between beighboring lambda windows is crucial for obtaining reliable free energy estimates. In our current protocols, the number of lambda windows is defined statically, before starting the calculation, not taking into account the difficulty of a transformation in an automated manner. This project will add tooling that selects the lambda schedule in an automated way, e.g. using a thermodynamic trailblazing algorithm (https://zenodo.org/records/3533199).
- Addition of automated lambda scheduling to a least one protocol's functionality, with working example.
- Significant experience with free energy calculations
@hannahbaumann, @atravitz
OpenFE is always looking to implement new protocols, including but not limited to:
- Binding Enthalpy Calculation
- LIE (linear interaction energy) Calculation (see this conversation for details)
- Transfer Free Energy Calculation (would replace this code from the Beckstein lab)
- MM/GBSA Calculation (see this conversation)
- A new protocol with complete tests and documentation
- Free Energy Calculations
- Familiarity with OpenMM
@hannahbaumann, @lalibay