Codon Optimization

We have moved beyond simple codon frequency tables. Our new algorithm prioritizes biological expression by focusing on:

• mRNA Stability
• Secondary Structure
• Ribosome Processivity

We have decoupled manufacturability parameters from the core optimization to focus primarily on protein expression. However, if the algorithm cannot find a manufacturable solution, it will automatically attempt to fix localized regions of extreme complexity. We do this while minimizing any impact on the expression profile.

Important: While we optimize for the best possible DNA composition, please remember that protein expression is multifactorial; DNA sequence is just one of many variables influencing your final yield.

Currently, we are unable to synthesize genes that score as Impossible. There are several factors that cause our algorithm to score a gene Impossible and the specific parameters are unique to the gene sequence. Our algorithm is proprietary and was developed using machine learning. In addition to machine learning, our algorithm does take into consideration the following factors:

• %GC between 25-65%
• Max homopolymer length < 10
• Low homology

A unique combination of the above is a reflection of the specific gene sequence score of Impossible.

Twist supports over 150 organisms, including most frequently used organisms such as E. coli, human, CHO, mouse and yeast.

We have moved beyond simple codon frequency tables. Our new algorithm prioritizes biological expression by focusing on:

• mRNA Stability
• Secondary Structure
• Ribosome Processivity

We have decoupled manufacturability parameters from the core optimization to focus primarily on protein expression. However, if the algorithm cannot find a manufacturable solution, it will automatically attempt to fix localized regions of extreme complexity. We do this while minimizing any impact on the expression profile.

Important: While we optimize for the best possible DNA composition, please remember that protein expression is multifactorial; DNA sequence is just one of many variables influencing your final yield.

Currently, we are unable to synthesize genes that score as Impossible. There are several factors that cause our algorithm to score a gene Impossible and the specific parameters are unique to the gene sequence. Our algorithm is proprietary and was developed using machine learning. In addition to machine learning, our algorithm does take into consideration the following factors:

• %GC between 25-65%
• Max homopolymer length < 10
• Low homology

A unique combination of the above is a reflection of the specific gene sequence score of Impossible.

Twist supports over 150 organisms, including most frequently used organisms such as E. coli, human, CHO, mouse and yeast.