OmegAMP Interactive Dashboard

Interactive companion to the OmegAMP paper -- a generative diffusion framework for antimicrobial peptide design.

Quick start

pip install modlamp

# Build locally
python build_dashboard.py --data-dir data --output docs/index.html

# Or build + encrypt with password
npm install -g staticrypt   # only needed for password protection
./deploy.sh --password YOUR_SECRET_HERE

Deploy with GitHub Actions (recommended)

The repository includes a workflow that builds the dashboard and deploys it to GitHub Pages automatically on every push to main.

Setup

1. Push this repo to GitHub.
2. Go to Settings > Pages > Source and select GitHub Actions.
3. (Optional) To enable password protection, go to Settings > Secrets and variables > Actions and add a repository secret named DASHBOARD_PASSWORD.
4. Push to main or trigger the workflow manually from the Actions tab.

The workflow will:

• Install Python + modlAMP
• Run build_dashboard.py to produce the HTML
• Encrypt with StatiCrypt if DASHBOARD_PASSWORD is set
• Deploy to GitHub Pages

If no password secret is configured the dashboard is published unprotected.

Manual deploy (alternative)

For environments without GitHub Actions, the deploy.sh script builds the dashboard into docs/ for branch-based Pages deployment:

./deploy.sh --password YOUR_SECRET_HERE
git add -A && git commit -m 'update dashboard' && git push

Then set Settings > Pages > Source to deploy from branch main, folder /docs.

Password protection

The dashboard is AES-encrypted with StatiCrypt. Without the password, the page source is an encrypted blob -- peptide sequences, MIC data, and all assay results are unreadable.

• Visitors see a login page and must enter the password
• "Remember me" saves a hashed credential in the browser for 30 days
• To change the password: update the DASHBOARD_PASSWORD secret and re-run the workflow

Views

Tab	Description
Activity	MIC across 20 strains + CC50/HC50/TI, sortable by family, group, or any column
Physicochemical Landscape	8 descriptors (charge, hydrophobicity, moment, aromaticity, ...) with strain-selectable MIC coloring
Membrane Activity	NPN/DiSC3(5) quadrant plots for membrane disruption profiling
Target Binding	LPS neutralization (dose-response +/- Ca2+) and DNA binding (deltaA), family-grouped bar charts
Structure	BeStSel secondary structure across 4 solvents
Safety	MIC vs CC50/HC50 scatter with TI diagonal lines, strain selector

Features

• Experiment group filters -- De novo, Inactive-to-active, LPS-binding, DNA-binding with All/None buttons
• Shared strain selector -- choose strains for MIC geometric mean across Safety and Physicochemical views
• Shared family/category filters across all tabs
• Search by peptide name
• Pin peptides -- persists across tab switches
• Brush-to-zoom on scatter plots
• Click column headers to sort in heatmap
• Export CSV -- download currently filtered data
• URL state -- shareable links that restore exact view

Input data

File	Rows	Required	Description
omegamp_reference_table.csv	217	yes	Peptide metadata
mic.csv	217	yes	MIC across 20 bacterial strains
cc50.csv	215	yes	Cytotoxicity CC50
hc50.csv	215	yes	Hemolysis HC50
disc.csv	196		DiSC3(5) membrane depolarization
npn.csv	196		NPN outer membrane permeabilization
bestsel.csv	190		BeStSel secondary structure
lps_binding.csv	577		LPS binding (BC displacement)
dna_binding.csv	111		DNA binding (deltaA)

Optional files are loaded if present; tabs with missing data render empty.

Strain panel

20 strains (15 Gram-, 5 Gram+), including 8 MDR clinical isolates:

Resistance	Strain	ATCC #
CRAB	A. baumannii	BAA-1605
CRE	E. coli	AIC222, BAA-3170
ESBL	K. pneumoniae	BAA-2342
FQR	P. aeruginosa	BAA-3197
MRSA	S. aureus	BAA-1556
VRE	E. faecalis, E. faecium	700802, 700221

Dependencies

• Python 3.8+ with modlAMP -- physicochemical descriptors
• Node.js with StatiCrypt -- password protection (optional)
• D3.js v7 -- loaded from CDN at runtime, no local install needed