Using Cradle

Overview

Cradle helps you engineer proteins better and faster through generative machine learning (ML). Cradle works for any modality and any property, customers have used the platform to optimize antibodies, enzymes, vaccines and more. You can start optimizing proteins with or without experimental data, no ML experience needed.

Using Cradle you generate sequences optimized for your experimental goals using models trained specifically on your protein. You test these sequences in the lab and import the assay data to Cradle. Based on this experimental data you train new models on Cradle that improve every round, learning more about your protein of interest every time.

You generate sequences use Cradle's ML through Tasks. Tasks can be configured using expertise on your protein of interest. You also have control over how you train models and select sequences with granular control of Tasks. To run Tasks to train models import your experimental to Cradle. You import data to Tables with schemas you provide, allowing for flexibility in representing the full context of your experimental data with consistent data quality.

Data

Data is central to your work with Cradle. Your experimental data is used as input in Tasks to generate sequences for lab testing. Every time you assay sequences in the lab you import your experimental data to Cradle. This helps the models trained through Tasks learn more about your protein of interest. Learn more in Data.

The quantity and quality of your data matters. To understand whether your data is suitable for machine learning, visit our data guidelines which cover minimum data requirements and how to structure different types of experimental readouts.

Tasks

You use Tasks to interact with Cradle's machine learning. Using Tasks you can train models, predict sequence performance, select libraries and more. The Task appropriate for your goal depends on your optimization stage and available data. Learn more in Tasks.

Reports

Use Reports to analyze Task outputs. Reports allow you to generate an analysis you can share with your team in an instant. Learn more in Reports.

Organizing your work

When using Cradle you can organize your work through:

• Workspaces. The home for all your data, projects and users for your organization. Data in the workspace will be private to your organization only. Learn more Workspaces.
• Projects. Within each workspace you can set up different projects. Each project works towards one optimization goal, such as developing a binder for a specific target or optimizing an enzyme for a specific catalysis. Learn more Projects.
• Rounds. To organize your project you can use Rounds. Each Round usually represents one cycle of generating sequences and testing them in the lab for a specific experimental goal. Rounds help you track which sequences have been tested and connect your computational work to lab workflows. Learn more in Rounds.

Get started

To get started with using Cradle, learn how you can set up your first Round.