Deep Origin

Deep Origin is a computational drug discovery platform that combines molecular physics with AI-driven acceleration to make small molecule drug discovery faster, smarter, and more affordable. Headquartered in South San Francisco, CA, the company serves research scientists, medicinal chemists, and drug discovery teams across the life sciences industry, enabling them to run complex in silico simulations and analyses without requiring deep computational expertise.

Backed by a world-class team of computational scientists, medicinal chemists, physicists, and biologists, Deep Origin transforms intricate simulation workflows into intuitive, accessible tools. The platform supports the full drug discovery pipeline — from target identification and hit discovery through to lead optimization — helping teams generate and evaluate more molecular candidates in silico before any synthesis takes place.

Key Products and Modules

• Balto AI (No-Code Interface): A no-code AI assistant for chemistry, docking, and drug discovery. Users can retrieve and analyze data from major databases including PDB, ChEMBL, UniProt, PubChem, BindingDB, and patents, as well as extract insights from PDFs and convert between SMILES, CAS, and IUPAC identifiers.
• Molecule Profiling via Balto: Predict key drug-like properties such as LogP, LogS, and LogD, assess ADMET risks including hERG, CYP450, and Ames toxicity, and evaluate synthetic accessibility — all through a plain-English interface.
• Protein Investigation via Balto: Locate binding pockets, align structures, and prepare proteins for docking workflows without writing code.
• Docking via Balto: Perform fast, high-accuracy ligand docking optimized for speed and no-code simplicity, with interactive 2D and 3D visualization of molecular structures, docking poses, binding pockets, and structural comparisons.
• Code-Based / API Access: A code-first interface for drug discovery at scale, supporting docking, free energy perturbation, and molecular analysis via programmatic workflows.
• Deep Origin OS Beta: A newly announced operating system for computational research, currently available for free sign-up.

Core Computational Capabilities

• Protein Preparation: Clean structures, fill missing loops, remove water and ions, select chains, and export prepared models for downstream use.
• Ligand Analysis: Build molecules from SMILES, SDF, or chemical identifiers; minimize structures; calculate ADMET properties; compute maximum common substructure (MCS); and export to SDF, MOL, or PDB formats.
• Docking Workflows: Detect binding pockets, define custom docking boxes, run docking simulations, view poses and scores, and export results for further analysis.
• Free Energy Simulations (ABFE/RBFE): Run end-to-end absolute and relative binding free energy simulations with full parameter control and trajectory visualization.
• Interactive Visualization: Render 2D and 3D representations of proteins, ligands, docking poses, and free energy perturbation networks throughout the workflow.
• PDF Data Extraction: A proprietary full-molecule data extractor for PDFs with reported 98% accuracy, enabling teams to mine literature and patent data efficiently.

Platform Differentiators and Notable Details

• Physics-Driven Accuracy: Simulations are grounded in molecular physics rather than pattern matching alone, providing results that researchers can trust for novel targets.
• AI Acceleration: Machine learning models are built to handle novel targets outside of existing literature, extending the reach of computational methods.
• Enterprise Scale: Infrastructure is designed to process millions of molecules, scaling with the ambitions of large research organizations.
• Speed: Workflows that previously took weeks can now be completed in hours, dramatically accelerating discovery timelines.
• Security and Compliance: SOC-2 compliant infrastructure with end-to-end encryption ensures that proprietary research data remains secure.
• ARPA-H Contract: Deep Origin is leading an up to $31.7M ARPA-H contract award to build in silico models intended to replace animal testing.
• Published Science: Research contributions span AI agents for drug discovery (arXiv), EGFR resistance mutations (npj Precision Oncology), neural network free energy methods, and multiscale modeling pipelines (ACS NMCS).
• Integrations and Data Sources: Native connectivity to PDB, ChEMBL, UniProt, PubChem, BindingDB, and patent databases.

Deep Origin is purpose-built for modern drug discovery teams seeking to leave no development candidate behind. By democratizing access to cutting-edge computational tools through both no-code and API-based interfaces, the platform enables scientists at every level to simulate biology, evaluate molecules faster, and advance more promising candidates toward the clinic.