For many molecular modelers, an all-too-common bottleneck is transforming a protein sequence into a reliable 3D structure. Whether you’re designing a new protein or studying the function of an uncharacterized one, waiting on complex installations or manually preparing jobs can slow down your workflow. Fortunately, SAMSON offers direct access to AlphaFold-2 in the cloud—no command-line or local setup required.
In this post, we’ll walk you through how to run AlphaFold-2 predictions entirely within SAMSON’s intuitive interface, letting you move quickly from sequence to structure with minimal manual configuration.
Why structure prediction still feels hard
While AlphaFold-2 has revolutionized structure prediction by providing highly accurate models, using it outside pre-packaged services can still be tricky. You often need to install specific tools, manage large databases, and navigate command-line scripts. On top of that, securing the required GPU compute—especially powerful ones like NVIDIA A100—can be expensive or out of reach for many individual users or small labs.
How SAMSON simplifies AlphaFold-2 predictions
The Biomolecular Structure Prediction extension in SAMSON integrates AlphaFold-2 directly into its interface. Predictions run on secure, high-performance cloud servers, which means you don’t have to worry about local hardware specs or software compatibility.
Steps to launch an AlphaFold-2 prediction in SAMSON:
- Go to Home > Predict in the SAMSON interface.
- Select the AlphaFold-2 prediction service.
- Upload one or more FASTA files containing your target sequences.
- Choose the model (monomer, multimer, etc.) and the multiple sequence alignment (MSA) database.
- Click Start prediction.
Once submitted, jobs are securely sent to cloud servers and you can track their status via Interface > Cloud jobs or your SAMSON Connect account.
Visualizing confidence with pLDDT scores
When results are loaded, SAMSON automatically colorizes models using pLDDT values (if present), helping you assess structural confidence at a glance. It’s a quick and helpful way to spot reliable domains versus uncertain regions before moving on to analysis or further modeling.
A few things to keep in mind
- Predictions require computing credits and are performed on cloud machines optimized for this task, including A100 GPU instances.
- You can purchase computing credits or contact the SAMSON team to request them.
- If you publish results based on AlphaFold predictions done via SAMSON, you should cite the original AlphaFold paper.
By embedding AlphaFold-2 into a user-friendly, cloud-enabled platform, SAMSON makes it far easier to integrate structure prediction into regular modeling workflows. You can free yourself from backend setups and focus on what matters: interpreting your models and designing the next experiment.
To learn more and explore Boltz-2 and Chai-1 prediction services as well, visit the full documentation page: https://documentation.samson-connect.net/tutorials/bsp/bsp/
SAMSON and all SAMSON Extensions are free for non-commercial use. You can get SAMSON here.