Predicting protein structures has long been a time-consuming challenge. Even with the advent of deep learning models like AlphaFold-2, integrating such powerful algorithms into your workflow can feel daunting.
But what if you could launch an AlphaFold-2 structure prediction directly from your molecular modeling software, with just a few clicks?
This post shows you how to do exactly that using SAMSON and the Biomolecular Structure Prediction extension. Whether you’re exploring new proteins or analyzing variants without known structures, this workflow can help you get reliable structure predictions quickly without leaving your modeling environment.
Why predict protein structures in SAMSON?
Instead of switching between tools, downloading files, or setting up scripts, you can work directly in SAMSON — a modeling environment built for molecular researchers. AlphaFold-2 predictions are launched from the interface and executed in the cloud, letting you focus on your science, not the setup.
How to use AlphaFold-2 in SAMSON
- In SAMSON, go to Home > Predict.
- Choose AlphaFold-2 as the prediction service.
- Upload one or more FASTA files.
- Select the model: monomer, multimer, etc., and choose the database for multiple sequence alignment (MSA).
- Click Start prediction. Your job runs in the cloud and typically finishes within minutes.
Results will appear under Interface > Cloud Jobs or in your SAMSON Connect job list.
Viewing confidence scores
Once your prediction loads in SAMSON, the structure is automatically colorized based on pLDDT values, a per-residue confidence metric. This helps you visually assess which parts of the model are predicted with high confidence and which are more uncertain.
Resources and costs
AlphaFold-2 predictions require computing credits and run on cloud machines with varying performance, including high-end A100 GPUs. If you’re running multiple predictions or longer sequences, this flexibility can significantly reduce your waiting time.
Contact the team or purchase credits to start.
When to use AlphaFold-2
- If a structure for your protein isn’t available in PDB.
- To explore structures of protein variants before synthesis or expression.
- To quickly visualize candidates in docking studies or interaction modeling.
Remember that when using the AlphaFold-2 service in research, it’s important to cite the appropriate publication: Jumper et al., Nature (2021).
For full details and advanced options (e.g., multimers, alignment databases), visit the full documentation:
https://documentation.samson-connect.net/tutorials/bsp/bsp/
SAMSON and all SAMSON Extensions are free for non-commercial use. You can download SAMSON at https://www.samson-connect.net.