How to Use Caliby Online
Try Caliby
Commercially Available Online Web Server
Caliby: Ensemble-Conditioned Protein Sequence Design
Caliby, a deep-learning-based model for ensemble-conditioned protein sequence design that successfully generates sequences for folding into a given target structure. Caliby addresses a major challenge in protein engineering: the difficulty of designing sequences that will fold into non-idealized backbones that are often out of reach for current models. The model hypothesizes that deep-learning models often struggle because they unintentionally learn non-structural signals, leading Caliby to focus on generating sequences that are generally generalizable and structurally plausible.
How Caliby Works
Caliby is a generative model that conditions its sequence design on structural ensembles rather than a single static structure:
Ensemble Conditioning: The model utilizes an ensemble approach (e.g., SolubleCaliby ensemble or SolubleCaliby) that prioritizes learning broadly generalizable structural signals over non-structural biases like phylogenetic relatedness.
Superior Accuracy: Caliby achieves superior performance over existing models (like SMPNN) in designing high-quality binders. The resulting designs show high predicted quality, with excellent predicted Binder PLDDT (around 90–100) and low Binder RMSD.
Target Applications: The model has been successfully applied to design binders for therapeutically relevant targets, including PD-L1, IFNAR2, CD45, and SpCas9.
Inference Refinement: The design process can involve an optimization step during inference, such as a rewind step (e.g., 150 rewind steps), to further refine the sequence and improve the predicted structure.
What is Tamarind Bio?
Tamarind Bio is a pioneering no-code bioinformatics platform built to democratize access to powerful computational tools for life scientists and researchers. Recognizing that many cutting-edge machine learning models are often difficult to deploy and use, Tamarind provides an intuitive, web-based environment that completely abstracts away the complexities of high-performance computing, software dependencies, and command-line interfaces.
The platform is designed provide easy access to biologists, chemists, and other researchers who may not have a background in programming or cloud infrastructure but want to run experimental models with their data. Key features include a user-friendly graphical interface for setting up and launching experiments, a robust API for integration into existing research pipelines, and an automated system for managing and scaling computational resources. By handling the technical heavy lifting, Tamarind empowers researchers to concentrate on their scientific questions and accelerate the pace of discovery.
Accelerating Discovery with Caliby on Tamarind Bio
Using Caliby on a platform like Tamarind would accelerate complex protein engineering and binder discovery campaigns:
Design for Non-Ideal Backbones: Researchers can design sequences for complex, non-idealized protein backbones that conventional inverse folding models struggle with, expanding the scope of engineerable scaffolds.
High-Quality Binder Generation: The platform could leverage Caliby's ensemble-conditioned approach to efficiently generate high-quality binder sequences for targets like PD-L1 and IFNAR2, providing a high-confidence set of leads for experimental validation.
Accessible Structural Optimization: Tamarind would handle the computationally intensive steps of running the deep learning model and the complex inference steps (like the rewind step), allowing researchers to focus on the biological relevance of the output sequences.
How to Use Caliby on Tamarind Bio
To leverage Caliby's power, a researcher could follow this streamlined workflow on Tamarind:
Access the Platform: Begin by logging in to the tamarind.bio website.
Select Caliby: From the list of available computational models, choose the Caliby tool.
Input a Target Structure: Provide the 3D structure (the backbone) of the protein you wish to design a sequence for.
Select Caliby Model: Choose one of the available Caliby models (e.g., the SolubleCaliby ensemble) that best matches your design goals.
Run Sequence Design: The platform runs Caliby, which performs its ensemble-conditioned sequence design to generate candidate amino acid sequences.
Acquire and Validate: The output provides a set of optimized sequences that are predicted to fold correctly into the target structure. These sequences can then be prioritized for experimental synthesis and testing.