Network Hamiltonian Model for CSH-CSSC system

This directory is the publication-facing subset of the NHMs project.

Structure

GitHub_Publication/
├── scripts/
│   ├── csh_cssc_1node_estimation_RestFiber.R
│   └── csh_cssc_1node_functions.R
├── plotting/
│   ├── AA_analysis_results.R
│   ├── CG_analysis_results.R
│   ├── NHM_analysis_plots.Rmd
│   ├── visualize_csh_cssc_networks.R
│   └── visualize_simulated_networks.R
├── results/
│   ├── README.md
│   ├── csh_cssc_1node_fit.Rdata
│   ├── csh_cssc_1node_fit_CG.Rdata
│   ├── csh_cssc_1node_fit_RestFiber.Rdata
│   └── final_AA_vs_CG_comparison_20250925_1857.dat
└── plots/
    └── Rplots.pdf

Scripts

• scripts/csh_cssc_1node_estimation_RestFiber.R: runs the restrained-fiber NHM fitting workflow and writes result files to results/.
• scripts/csh_cssc_1node_functions.R: helper functions used by the restrained-fiber NHM fitting script.
• plotting/AA_analysis_results.R: loads atomistic fit results and generates summary tables and GOF plots.
• plotting/CG_analysis_results.R: loads coarse-grained fit results and generates summary tables and GOF plots.
• plotting/NHM_analysis_plots.Rmd: main R Markdown driver for the bundled plotting workflow.
• plotting/visualize_csh_cssc_networks.R: plots molecular-dynamics network snapshots when the underlying simulation data are available locally.
• plotting/visualize_simulated_networks.R: plots simulated fitted networks if simulated-network result files are added separately.
• results/README.md: documents the structure of the bundled .Rdata result files and the comparison data file.

ERGM Terms

The restrained-fiber fitting script evaluates the following candidate ERGM/NHM terms. A baseline edges term and Krivitsky size offset are added automatically by the fitting code and are not listed in cand.

• kstar(2): two-star term; captures tendency for nodes to form multiple ties.
• nodematch("Resname"): overall homophily by residue type, i.e. preference for same-type ties.
• nodematch("Resname", levels=I(resnames), diff=TRUE): type-specific homophily, separating same-type effects by residue class.
• nodefactor("Resname", levels=I(resnames)[2]): main-effect shift associated with the second residue type, CSSC.
• isolates: counts degree-0 nodes.
• dimers: counts isolated two-node connected components.
• degree(0, ...), degree(1, ...), degree(2, ...), degree(3, ...): residue-specific degree terms; these track how often nodes of a given residue type have degree 0, 1, 2, or 3.
• gwdegree(0.25, fixed=TRUE), gwdegree(0.5, fixed=TRUE), gwdegree(1, fixed=TRUE), gwdegree(2, fixed=TRUE), gwdegree(3, fixed=TRUE): geometrically weighted degree terms; these summarize overall degree heterogeneity while downweighting very high degrees.
• components: counts connected components.
• compsizesum(pow=2): emphasizes larger connected components by summing squared component sizes.
• esp(0), esp(1), esp(2): edgewise shared partner terms; these count edges whose endpoints share 0, 1, or 2 common neighbors.
• nsp(1), nsp(2): non-edgewise shared partner terms; these count nonadjacent node pairs sharing 1 or 2 common neighbors.
• gwesp(0.05, fixed=TRUE), gwesp(0.25, fixed=TRUE), gwesp(0.5, fixed=TRUE), gwesp(0.75, fixed=TRUE), gwesp(1, fixed=TRUE), gwesp(2, fixed=TRUE): geometrically weighted edgewise shared partner terms; these summarize triadic closure / clustering across shared-partner counts.
• degcor: degree-correlation term; captures dependence between the degrees of linked nodes.

Results

See results/README.md for the .Rdata directory tree and object structure.

Data Note

This bundle includes analysis and plotting code, selected result files, and plot outputs. Large raw simulation data are not included.