Improving Codon Optimization with Machine Learning

Overview

Protein sequences can only contain 20 different amino acids. However, these amino acids are encoded by 61 sense codons. Using synonymous codons, while not affecting the amino acid sequence and protein structure, can have an impact on the expression of the protein. Codon optimization focuses on finding amino acid sequences with the highest expression, thus playing a crucial role in biotechnology and synthetic biology. Traditional methods rely on codon usage tables which suffer from limitations in accuracy and adaptability across diverse organisms due to usage bias. This study proposes a novel approach to codon optimization by utilizing a long short-term memory (LSTM) machine learning model. The LSTM model was trained on DNA sequences to learn codon usage bias and other patterns not considered by traditional codon optimization techniques. The LSTM model takes in an amino acid sequence and generates the corresponding optimized DNA sequence.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• AJAS None projects
• Every AJAS 2025 project