GeoBiologics ¶

Meet GeoBiologics, an AI-powered one-stop platform for protein design and optimization.

Catering to drug designers at pharmaceutical companies and research institutions, GeoBiologics harnesses the power of BioGeometry’s advanced artificial intelligence algorithms and presents them in an intuitive visual format for streamlined protein design.

Design goals¶

Intelligent	Efficient	Intuitive
GeoBiologics' intelligent predictions and generations are powered by BioGeometry’s proprietary AI algorithms and data.	GeoBiologics' streamlined workflow and powerful cloud computing capabilities allow you to design proteins faster than ever before.	GeoBiologics' intuitive visual interface makes it easy to design proteins, visualize results, and export data.

Quickstart¶

Capabilities¶

GeoBiologics is a one-stop, comprehensive platform for antibody design and optimization. It empowers a wide range of antibody-related tasks with AI algorithms, including:

Utils¶

Interface Visualization: Visualize a protein-protein interface within a structure file.
Structure Alignment: Align a set of structures to a target structure in a specified manner. Optionally, compare another part of the structure (e.g., calculate ligand RMSD).

Structure modeling¶

General protein structure prediction: Predict the structure of input protein sequence(s). Optionally, search for MSA and structural templates as additional inputs.
Antibody structure prediction: Predict structures of antibody variable domains.
Antibody-antigen docking: Predict the structure of an antibody-antigen complex given the antibody and antigen sequences.

Affinity Optimization¶

Affinity optimization: Optimize the affinity of a ligand protein to a target protein by saturated mutagenesis and ranking.

Antibody Design¶

Antibody screening and optimization: Comprehensively evaluate and optimize your lead antibodies with powerful and specialized AI assistants.
De novo design of target-specific CDR library: Design a CDR library de novo given a starting antibody and a target epitope.
Humanness prediction (legacy): Predict the humanness of the given antibody sequence(s), defined as the average occurrence of the 9-mer peptide fragments among OAS human assays.

Developability prediction (legacy): Predicts various antibody developability properties from sequence, including solubility, stability, expression, polyspecificity and immunogenicity.
Humanization (legacy): Modify an antibody to be similar to human antibodies, potentially reducing its immunogenicity in humans.

Protein Design¶

Protein inverse folding: Given a protein/antibody structure, generate a sequence that is compatible with the structure.
Enzyme optimization: Optimize the properties (activity, pH stability, thermostability, and stereoselectivity) of an enzyme given its sequence and the reaction it catalyzes.
Enzyme optimization (training): Fine-tune our protein foundation model with your data to predict properties of your enzyme with higher accuracy.
Enzyme mining: Identify high-quality starting enzymes that can efficiently catalyze specific reactions from a rich library of enzymes.

About BioGeometry¶

BioGeometry is a digital biology pioneer that specializes in applying generative AI to protein design. By constructing a data-driven bridge between protein sequence, structure and function, we are committed to generating proteins de novo with programable properties for biomedicine and bio-manufacturing.

FAQ¶

I still have questions. How can I get help?¶

Feel free to reach out to us either by submitting a feedback, or by sending an email to system@biogeom.com.

To submit a general feedback, hover on your avatar on the top right (if you are in the project editor) / bottom left (if you are in the control panel) corner of the page, click on "Share your feedback" and fill in your question there.
To submit a feedback specific to a job, go to the job result page, and fill in the feedback form at the bottom of the page.