An unsupervised learning and robust surface curvature measure for surface complementarity - Abhijit Gupta, Arnab Mukherjee
- The src directory contains jupyter notebook files for running model comparison - where we compare our model with coleman's model using analytical dataset comprising of perturbed points on spheres with known radius and curvature (curvature = 1/R)
- The other jupyter notebook contains demo for Protein-inhibitor system, which showcases our curvature based surface complementarity function.
- The utils directory has two programs - hypersphere.py, which has our core surface curvature algorithm, and read_msms.py, which is used for reading MSMS files.
- For generating SES surface (solvent excluded molecular surface), please install MSMS program or use Chimera, which has built in utility of generating SES surface, saved as dot molecular surface.
- Python libraries - (i) NumPy (ii) Scipy (iii) matplotlib (iv) regex (v) unittest (vi) pandas (vii) Biopython (viii) PyMesh
python -m pip install numpy scipy matplotlib regex pandas biopython pymesh jupyter signac signac-flow