EnCodon / DeCodon

EnCodon and DeCodon are components of a codon-aware foundation model framework for exploring and manipulating the sequence-function landscape of coding RNAs. EnCodon enables efficient codon embedding and sequence representation, while DeCodon allows for controlled sequence generation and decoding, offering a powerful toolset for understanding codon usage, translation regulation, and engineering novel coding sequences.

Click here to access EnCodon and DeCodon

Tool Features

Codon-based tokenizer and foundation model pretraining across taxa

Flexible interface for sequence embedding, probing, and downstream modeling

Sequence generation and optimization via DeCodon

Designed for plug-and-play use with PyTorch and HuggingFace ecosystems

Supports applications in codon optimization, translation modeling, and design