Monte Carlo peptide search

This is my fork of a Monte Carlo tree search implementation for self-assembling peptide design, adapted to explore aggregating peptide sequences. The original work pairs the search with a random-forest surrogate, so every candidate doesn't need its own molecular dynamics simulation.

The demo searches through tripeptides using precomputed scores for all 8,000 sequences. It's a way to explore which candidates are worth a closer look before committing to more expensive evaluations.