First Ascent by
Jorge Carmona
The Chameleon Traverse
Cheminformatics / Drug Permeability
The Proposed Route
A technical traverse across the conformational landscape of cyclic peptides โ molecules that change shape depending on their environment (aqueous vs. membrane). By computing 3D descriptors at two dielectric constants and correlating the difference (ฮ) with experimental permeability, the climber attempts to quantify the 'chameleonic potential' that 2D fingerprints completely miss. This route connects directly to Jorge's thesis DEL platform.
๐ง The Crux
MMFF94 force field minimization in implicit solvent is a crude approximation for conformational sampling โ the chameleonic effect may require higher-level methods (GB/SA, explicit solvent) to capture. The Tier 2 cross-check with OMEGA + OpenMM on reference compounds is essential to validate whether Tier 1 ฮ descriptors are meaningful. Assay heterogeneity in CycPeptMPDB is a known issue.
โ ๏ธ Pre-Climb Checklist
โ CycPeptMPDB freely downloadable. โ Academic OpenEye license for OMEGA. โ ๏ธ Lock in molecule set and confirm PAMPA subset before starting Tier 1. โ ๏ธ DEL stretch requires PI approval โ exclude from minimum viable deliverable. โ GitHub README with single-command run is a great deliverable standard.
Guidance
- Be upfront about MMFF94 limitations โ Tier 2 cross-check is your insurance
- UMAP colored by LogPexp with CycloA overlay = most compelling figure โ make it publication-quality
- Weak correlation is still a result โ means 3D MMFF descriptors insufficient
Source proposal: Carmona_Jorge_Final_Proposal_CyclicPeptide.docx
CHEM 169/269 ยท Applied AI & Machine Learning for Biochemistry