Biosphere
Docking, molecular dynamics, and free-energy workflows in one workspace. Run with managed cloud, hybrid compute, or advanced local GPU when your lab needs it.
Managed cloud is the default. Hybrid is the recommended lab path. Local GPU remains available as an advanced mode.
Institutions
Simulations
Researchers
Trusted by researchers at 50+ institutions worldwide
Three steps to discovery
Go from a protein target to actionable drug discovery results in hours, not weeks.
Upload Your Target
Upload a protein structure (PDB) or provide a target identifier linked to an external structure backend. The validated local path uses explicit receptor structures plus QC and preparation.
Configure & Launch
Select pipeline phases, set simulation parameters, and add ligands via SMILES, PubChem, or ChEMBL search. One click to launch your full workflow.
Get Results
Receive docking poses, MD trajectories, binding energies, and auditable scientific reports. Analyze in our interactive 3D viewer or export in standard formats.
End-to-end automated workflow
Each phase builds on the previous - fully automated with real-time monitoring.
Structure Prep
Experimental receptor QC and preparation
Binding Site
Pocket detection, druggability assessment
Ligand Prep
Ligand standardization and topology handoff
Docking
AutoDock Vina docking
MD Prep
System building, solvation, equilibration
MD Simulation
GPU-accelerated GROMACS, 3 replicas
Analysis
RMSD, RMSF, H-bond analysis
Free Energy
MM-PBSA binding affinity
Everything you need for drug discovery
Production-grade tools used by pharmaceutical researchers and academic labs.
Auditable Workflow Controls
Track every phase, checkpoint, and artifact with explicit approvals, export-ready records, and workflow safeguards designed for reproducible scientific runs.
Molecular Dynamics
GPU-accelerated GROMACS simulations with 3 replica runs. Up to 200ns per replica with real-time energy monitoring.
Free Energy Calculations
MM-PBSA binding affinity with per-residue decomposition. Quantitative insights for lead optimization.
Real-time Monitoring
Live pipeline progress via WebSocket. Track each phase, process, and replica as it runs.
Interactive 3D Viewer
NGL-based molecular visualization. Rotate, zoom, and analyze docking poses and MD trajectories.
Export & Share
PDB, SDF, CSV, and auditable scientific reports. Share results with collaborators instantly.
Interactive structure visualization
Rotate, zoom, and explore protein structures and ligand poses with our WebGL-powered NGL viewer.
Loading 3D Viewer...
Initializing WebGL context
Transparent compute pricing
Managed cloud is the default, hybrid is the recommended lab option, and local GPU remains an advanced mode for teams that already maintain their own scientific stack.
Compute Profiles
Local GPU is available as advanced lab modeRunPod-class managed cloud execution for routine production studies
Priority managed cloud for larger systems and urgent turnaround
Hybrid is the recommended split for labs; local-only remains an advanced mode for teams with their own scientific stack
Credit Packs
Pre-purchase compute credits for managed, hybrid, or advanced local execution so your team can budget controlled runs and scale without surprise infrastructure costs.
Entry pack for validation or setup runs
Save $30 vs pay-as-you-go credit pricing
Save $150 vs pay-as-you-go credit pricing
How It Works
Upload
Provide your protein structure and ligand
Choose Compute
Start with managed cloud, choose hybrid if you have a lab GPU, and use local-only only in advanced setups
Run
Launch the workflow and monitor phases, replicas, and checkpoints
Bill Precisely
Managed runs bill GPU-hours; local mode bills lightweight orchestration only
Running high-volume simulations? Contact us for enterprise pricing
Trusted by leading researchers
"Biosphere cut our docking-to-MD workflow from 3 weeks to 8 hours. The explicit receptor-preparation path and standardized Vina-to-MD handoff saved us countless hours of manual setup."
Dr. Sarah Chen
Stanford University
Structural Biology"The MM-PBSA module gives us replica-level free energy summaries with per-residue decomposition and clear provenance. It made our internal review process much faster."
Prof. James Okafor
University of Oxford
Computational Chemistry"As a small lab with limited GPU resources, the managed-versus-local compute options are exactly what we needed. We can validate locally first, then burst to cloud only when it is worth paying for."
Dr. Mei-Lin Zhang
MIT
Drug DesignWhat researchers say about Biosphere
Trusted by 500+ researchers across 50 institutions for drug discovery simulations.
Ready to accelerate your drug discovery?
Launch docking, MD, and free-energy analysis from one workspace with a clearer path from setup to results.
Managed cloud by default. Hybrid for labs. Advanced local GPU available when needed.
Biosphere Documentation
Practical guidance for setup, workflow usage, pricing, API access, and infrastructure configuration.
Start your first simulation
From sign-up to GPU results in under 5 minutes. No installation required.
Sign up with your email and password. Your account becomes active immediately and the workspace is available at once.
- Click "Get Started" on the landing page
- Enter your name, email, and password
- Open the workspace and prepare your first target
Purchase credits when you are ready for paid compute. Credits do not expire and work across managed, hybrid, and advanced local runs.
- Open the Pricing section
- Choose a credit pack that fits your study size
- Complete secure payment via Stripe
Provide a protein target, add a ligand, choose the workflow settings, then start the run from the left panel.
- Upload a PDB file or enter a UniProt ID
- Add ligand via SMILES or SDF upload
- Click "Start Workflow" and monitor the phases in real time
8-phase automated workflow
Each phase builds on the previous output. Select all phases for a complete CADD workflow or pick individual phases as needed.
Prepares your protein structure for simulation by running QC validation, adding missing atoms, and standardizing the receptor for downstream docking and MD. Sequence-only targets require a connected external structure-generation backend.
PDB file, prepared receptor structure, or a target identifier with an external structure backend
Cleaned PDB, QC report, preparation log, receptor handoff files
Identifies and ranks potential ligand-binding pockets using geometric and energy-based algorithms. Each pocket is scored for druggability, volume, hydrophobicity, and residue conservation.
Prepared protein structure from Phase 1
Pocket coordinates, druggability scores, key residues list, grid box parameters
Standardizes ligand chemistry into docking-ready and MD-ready handoff files, records protonation/geometry preparation, and exposes when external cheminformatics filters or parameter backends are required.
SMILES string, SDF file, or PubChem search
Docking-ready ligand, MOL2 handoff for ligand topology, preparation scope report
Runs standardized AutoDock Vina docking with explicit pose ranking, selection provenance, and clean handoff into MD preparation. Downstream free-energy analysis is performed only when the required backend is available.
Prepared structure, binding site grid, ligand parameters
Docking poses (PDB/PDBQT), Vina scores, pose metadata, interaction maps
Builds the complete simulation system by placing the protein-ligand complex in a solvated box, adding ions for neutrality, assigning CHARMM36m force field, and running energy minimization + equilibration (NVT/NPT).
Best docking pose, protein topology, ligand parameters
Solvated system (.gro), topology (.top), MDP files, equilibration logs
Runs GPU-accelerated GROMACS molecular dynamics with 3 independent replicas for statistical robustness. Produces trajectory data (coordinates, velocities, energies) at configurable time steps with real-time energy monitoring.
Equilibrated system, GROMACS topology, MDP parameters
Trajectory files (.xtc/.trr), energy files (.edr), log files, checkpoint files
Analyzes trajectory data to compute RMSD (convergence), RMSF (flexibility), hydrogen bond occupancy, radius of gyration, secondary structure, and principal component analysis (PCA) for essential dynamics.
Trajectory files from Phase 6
RMSD/RMSF plots, H-bond analysis, PCA projections, summary statistics
Computes binding free energies using the MM-PBSA method across all 3 replicas. Provides per-residue energy decomposition to identify key binding contributions, with convergence analysis and experimental comparison.
Trajectory files, topology from Phase 6
Delta G values, energy decomposition, per-residue breakdown, convergence data, final report
Understanding compute-credit pricing
Transparent pricing across managed cloud, recommended hybrid execution, and advanced local GPU mode. Credits map directly to billed compute value with no hidden lock-in.
How billing works
Biosphere now uses compute credits instead of nanosecond bundles. Managed cloud is the default path, hybrid is the recommended practical option for labs, and local GPU remains an advanced mode for teams already maintaining the required scientific stack. Credits never expire and can be reused across jobs.
What local GPU pricing means
In advanced local mode, you are not paying Biosphere for your lab electricity, your owned GPU, or raw hardware time. The platform charges for orchestration, provenance tracking, workflow automation, result syncing, and optional managed fallback when hybrid execution is enabled.
Validation remains a scientific step
Trust local or hybrid results only after reviewing RMSD stability, energy convergence, and replica consistency. The workflow exposes these outputs later so researchers can judge run quality instead of treating GPU completion as proof by itself.
Compute profile comparison
Credit pack pricing
| Pack | Price | Credits | Notes |
|---|---|---|---|
| 100 credits | $100 | 100 | Starter validation pack |
| 315 creditsPopular | $285 | 315 | Most popular (+15 bonus) |
| 825 credits | $675 | 825 | Best value (+75 bonus) |
How we keep costs low
We leverage specialized GPU providers and durable workers instead of heavyweight general cloud stacks. Managed runs stay hosted, while labs with their own hardware can usually start with hybrid execution and move to local-only later if they already manage the scientific tooling themselves.
REST API Preview
Programmatic access to Biosphere functionality. Full API docs with authentication coming soon.
| Method | Endpoint |
|---|---|
| GET | /api/demo/presets |
| POST | /api/demo/simulate |
| POST | /api/payments/checkout |
| GET | /api/payments/balance |
| GET | /api/payments/history |
Frequently asked questions
Everything you need to know about using Biosphere. Can't find your answer? Reach out through the contact options on the platform.
Backend Setup Guide
Configuration instructions for deploying and managing Biosphere infrastructure.
Railway web app
Hosts the main Next.js interface, API routes, auth, billing, and job orchestration.
Neon database
Stores users, jobs, ratings, messages, and workflow state through Prisma/PostgreSQL.
Render relay
Keeps realtime job updates flowing from the platform to the browser through the websocket relay.
Managed or local workers
Run Vina, GROMACS, ACPYPE, fpocket, and optional AmberTools, then push artifacts and progress back.
- 1Create an account at vast.ai and complete identity verification
- 2Add a billing method (credit card or bank transfer)
- 3Generate an API key from Account Settings -> API Keys
- 4Configure the worker host with Ubuntu 22.04, CUDA 12.x, Python 3.11+, and at least 80 GB of fast local disk
- 5Install pipeline tools natively on the worker host: GROMACS, AutoDock Vina, OpenBabel, fpocket, and optional AmberTools for MM-PBSA.
- 6Run a manual preflight before production use: verify `nvidia-smi`, `gmx --version`, Vina, OpenBabel, fpocket, and storage paths. A dedicated doctor-style command is not exposed in the main product yet.
- 1Create an account at runpod.io and add billing
- 2Configure serverless GPU endpoints for each pipeline tool
- 3Set up worker runtimes: provision hosts or images with GROMACS, AutoDock Vina, OpenBabel, fpocket, and optional AmberTools installed.
- 1Each launched job becomes a durable workflow run that can be claimed by managed or local workers
- 2Workers push phase and process progress, checkpoint snapshots, and results back into the orchestrator for resume and auditing
- 3Results and artifacts can be persisted in S3/R2-compatible object storage for durability and retrieval
