Sequence generation

Overview

Once you have created a round, you can start with generating sequences. Before the models start generating for you, you will configure what you want the design to look like. You will be able to pick the size of the library, choose a starting point, and configure mutations.

When Cradle has generated sequences for you, you can view them in a generation report.

Starting a configuration

Once you have created a round, you can start configuring a generation by clicking on Configure generation.

Not working? Check if you have met the requirements.

• You have added at least one assay.

• You have added a primary objective based on the assay you have added.

• You have created a round.

Configuring generation

When choosing how you want to design your sequences, go through the following steps:

1

Experimental capacity

Choose the size of the library that Cradle will generate for you. Cradle models choose an optimal mix between:

Exploration. This allows the model to search for novel mutations (not yet included in your uploaded sequences), reducing risks in the optimization campaign and learning more about the context of your protein of interest.

Exploitation. This focuses the model on creating sequences with the most performance and helps you achieve your experimental goals.

You can choose a library size of 48, 96, 384, or a custom number of candidates.

2

Design starting point

Pick the round template sequence that will be used as the starting point of the sequence generation. You can choose:

• To use your project base sequence.

• Or select a sequence from your project data. If you choose this option, you will need to select a sequence from the uploaded data. Click on Done once you have selected the sequence you want to start with.

Pick a high-quality sequence as your starting point to increase the quality of the sequences generated by the models.

3

Number of mutations

Select the minimum and maximum number of mutations that you want to see in the generated sequences. Cradle generates sequences with varying mutation loads to balance novelty, performance, and design risk. You can control the number of mutations to meet specific project needs.

Increasing the minimum number of mutations will lead to more novel sequence candidates.

The maximum number of mutations can be viewed as the level of design risk you're willing to take. More mutations mean higher risk, but could also lead to better results.

4

Blocked mutations

If you want to prevent changes in specific regions of the protein, you can block specific mutations from appearing in the generated sequences by setting Blocked mutations. Here's how to do it:

• Set blocking rules: Choose one or all amino acids that you want to block from appearing in specific regions or across the entire protein.

• Define a region: Specify where these blocked mutations should apply. To block mutations in specific areas (e.g., core or surface), navigate to Manage Regions, create a new region, and assign the blocking rules to it.

Downloading & viewing generated sequences

When Cradle finishes generating sequences, you will receive a notification.

You can download the sequences and view a generation report to understand where the model made mutations.

Testing in the lab

After downloading the sequences, you can test them in the lab. The library is optimized to contain sequences that exploit and explore the protein space of interest. Therefore, to maximize learning and performance, you need to test the entire library generated by Cradle.

Once you have tested the library, you upload your data back to Cradle. Based on what you learned from this round, you can configure a new round to get closer to your experimental goals.