MS2Int Web Server

Web interface for MS2Int peptide MS/MS spectrum prediction.

Features

• Single Mode: Predict spectrum for one peptide — returns annotated PNG + ion table
• Batch Mode: Upload CSV/TSV → async job queue → download H5 results
• Pixel Art UI: NES-inspired design with dark mode support

Quick Start

# 1. Activate conda environment
conda activate mamba

# 2. Install frontend dependencies (first time only)
cd webserver/frontend && npm install && cd ..

# 3. Start both servers
bash start.sh
# Options: --gpu 0 --port 8000 --frontend-port 5173

Open http://localhost:5173 in your browser.

Architecture

webserver/
├── backend/          # FastAPI (Python)
│   ├── app.py        # Main app + API routes
│   ├── config.py     # Paths, GPU, model config
│   ├── predictor.py  # Model loading + inference
│   ├── ion_labels.py # Ion annotation + m/z calc
│   ├── spectrum_render.py  # spectrum_utils PNG
│   ├── schemas.py    # Pydantic models
│   └── job_manager.py     # Batch job queue
├── frontend/         # React + Vite + TailwindCSS v4
│   └── src/
│       ├── components/    # UI components
│       ├── api.ts         # API client
│       └── index.css      # Pixel art design tokens
├── start.sh          # One-click launcher
└── PLAN.md           # Full development plan

API Endpoints

Method	Path	Description
GET	/api/health	Health check + model status
POST	/api/predict	Single peptide prediction
GET	/api/supported-modifications	List supported PTMs
POST	/api/jobs/submit	Submit batch CSV/TSV
GET	/api/jobs/{job_id}	Query job progress
GET	/api/jobs/{job_id}/download	Download H5 results
GET	/api/jobs	List all jobs

Single Prediction Example

curl -X POST http://localhost:8000/api/predict \
  -H "Content-Type: application/json" \
  -d '{"sequence":"PEPTIDEK","charge":2,"collision_energy":30,"fragmentation":"HCD"}'

Batch CSV Format

Sequence,Charge,collision_energy,Fragmentation
PEPTIDEK,2,30,HCD
[Acetyl]-ALLS[Phospho]LATHK,3,25,HCD

Required columns: Sequence, Charge, collision_energy, Fragmentation

Supported Parameters

• Charges: 1–6
• Collision Energies: 10, 20, 23, 25, 26, 27, 28, 29, 30, 35, 40, 42
• Fragmentation: HCD, CID
• Max peptide length: 30 amino acids
• Modifications: M[Oxidation], S[Phospho], T[Phospho], C[Carbamidomethyl], [Acetyl]-, K[Acetyl], K[GG], and more

Requirements

• Python 3.10+ with mamba_ssm, torch, fastapi, spectrum-utils
• Node.js 18+ for frontend
• CUDA GPU (~50MB VRAM for model)