Predicting protein structures with high accuracy can be a complex and time-consuming task, especially when setting up local tools like AlphaFold. Many researchers spend hours installing dependencies, managing GPUs, and dealing with input/output incompatibilities. But what if you could bypass all of that and use a cloud-based, GPU-accelerated service with just a few clicks?
This is where SAMSON’s Biomolecular Structure Prediction extension and its integration with AlphaFold-2 come in. In this post, we walk through how you can use AlphaFold-2 directly through SAMSON, predict monomers or multimers, and visualize prediction confidence – no local setup needed.
Why use AlphaFold-2 in SAMSON?
AlphaFold-2 has set a new standard in protein structure prediction. However, running it locally requires significant hardware resources and technical know-how. In contrast, SAMSON offers:
- Cloud-based predictions that use A100 GPUs
- An intuitive interface for uploading FASTA files
- Automatic coloring of predicted structures based on pLDDT confidence scores
- Support for multiple models (monomers, multimers)
- Monitoring tools via the SAMSON interface and SAMSON Connect
How to Get Started
To predict structures using AlphaFold-2 in SAMSON, simply follow these steps:
- In SAMSON, open Home > Predict.
- Select AlphaFold-2 as your prediction service.
- Upload your protein sequences in FASTA format.
- Choose the appropriate AlphaFold model — for instance, monomer or multimer.
- Select a database for multiple sequence alignment (MSA).
- Click Start prediction.
That’s it. Your job is now running in the cloud. You can track progress and view results through:
- Interface > Cloud jobs in SAMSON
- Your account on SAMSON Connect
Interpreting the Prediction
Once the structure loads in SAMSON, residues are colorized based on their per-residue confidence score (pLDDT), giving you insight into areas of high or low reliability. This visual feedback helps you quickly gauge which parts of the model are trustworthy and where caution is needed.
Compute Credits and Cost
Predictions consume cloud computing credits. Users can choose machine types based on desired performance and budget. Compute credits may be:
- Requested via email
- Purchased from the computing credits page
This flexibility makes it accessible to teams with different prediction needs and resources.
Publication Note
If you publish results obtained from AlphaFold predictions in SAMSON, make sure to cite the original AlphaFold paper:
This credits DeepMind’s contribution and maintains good scientific practice.
Conclusion
SAMSON offers an accessible, cloud-based interface to AlphaFold-2, allowing molecular modelers to focus on science instead of server maintenance. Whether you’re predicting the structure of a single protein or analyzing a complex, the integration in SAMSON simplifies the entire workflow from sequence to model.
To learn more, check the full documentation page.
SAMSON and all SAMSON Extensions are free for non-commercial use. You can download SAMSON at https://www.samson-connect.net.