GSoC 2025 - Project 6

[sunagparasu]

Hello,
My sincere apologies for starting a new discussion this late! My name is Sunag and I'm a masters student in bioinformatics at Saarland, Germany. During my degree, I have developed an interest in computer aided drug design and would like to contribute to the MDAnalysis library.
I have completed multiple ML/DL for drug discovery projects and have used fingerprints (morgan fingerprints, to be more accurate) within those projects and understand the importance of interaction fingerprints within molecular dynamics.
I believe we can make IFPs continuous by calculating the dipole moments of each interaction (this value would change as it is dependent on distance and angle between the two interacting atoms) and then either normalize these values or use a sigmoid function to re-scale this to fit a range between 0 and 1.
Another approach would be to use existing protein ligand interaction data to train ML/DL models to output a probability distribution curve for each interaction type.
I do understand this is an unpolished proposal, but I absolutely could provide a detailed proposal that would split the work into weekly based tasks if you'd be interested.
I also apologize for submitting a pre-proposal before starting a new discussion.