Many molecular modelers frequently wrestle with the challenge of predicting protein structures when experimental data is out of reach. Whether working on novel sequences or comparing predictions across homologs, the bottleneck often lies in accessing the right tools that balance ease of use, computational power, and flexibility.
If you’ve ever wondered how to efficiently turn a .fasta file into a 3D protein structure without worrying about local installation or hardware constraints, SAMSON offers a streamlined solution thanks to its integration with AlphaFold-2.
Predicting Structures with AlphaFold-2 in SAMSON
The Biomolecular Structure Prediction extension in SAMSON allows you to run AlphaFold-2 structure predictions directly from the GUI without having to manage dependencies or cloud APIs manually. Everything is integrated into a few simple steps.
Here’s how to start your prediction:
- Launch SAMSON and access the prediction tools via Home > Predict.
- Select the AlphaFold-2 service.
- Upload one or more
.fastafiles of sequences to model. - Choose a model type—monomer or multimer—depending on the nature of your input.
- Select a multiple sequence alignment (MSA) database to use.
- Click Start prediction.
That’s it. The job runs in the cloud, and there’s no need to worry about your local processing power. Predictions can be monitored through Interface > Cloud Jobs in SAMSON or via your user account online.
Why It Simplifies Prediction Workflows
Running AlphaFold-2 typically requires setting up a complex software stack with Docker, databases, and high-end GPUs. SAMSON abstracts that completely, letting modelers focus on the science instead of the infrastructure.
In addition, when you load the results in SAMSON, structures are automatically colorized by pLDDT (per-residue confidence), making visual interpretation of prediction quality easier.
What About Resources?
SAMSON allows you to pick between various cloud machines with different performance levels and computing credit costs. For intensive projects, high-performance A100 GPU instances are available.
You can request credits by emailing contact@samson-connect.net or purchase them through your SAMSON account.
How Should You Cite?
If you publish any findings based on AlphaFold predictions in SAMSON, be sure to cite the original AlphaFold-2 paper from Nature (2021). If you used the multimer model, include the relevant multimer paper as well.
Use Case Example
Imagine being able to test folding hypotheses of protein variants or visualize hypothetical candidates for synthetic biology applications directly in your modeling workflow. The speed and convenience make SAMSON’s AlphaFold integration a helpful companion tool for everyday structure modeling tasks.
Learn more about AlphaFold predictions and other structure prediction tools at the official SAMSON Biomolecular Structure Prediction documentation page.
SAMSON and all SAMSON Extensions are free for non-commercial use. To get SAMSON, visit https://www.samson-connect.net.