Overview
Modern models learn directly from sequence. We apply them to predict function, regulatory effects and new designs.
Salient features
- Sequence-to-function prediction
- Embeddings and function annotation
- Generative protein design
Analyses we offer
Deep learning for genomics (sequence-to-function)
Predicts regulatory activity and variant effects directly from DNA sequence.
Methods: borzoi, kipoi, alphagenome.
Protein language models & function prediction
Uses language-model embeddings to predict protein function and properties.
Methods: esm, ProtTrans, DeepFRI.
Protein design
Generates new protein backbones and sequences for a target structure or function.
Methods: RFdiffusion, ProteinMPNN.
What you receive
- A written report with methods, parameters and interpretation
- Processed data tables and publication-quality figures
- Scripts and settings so every result can be reproduced
Not sure which analysis fits your data? Contact us and we will advise.