Predicting protein structures has traditionally been a time-consuming and computationally intensive task. With the rise of deep learning approaches like AlphaFold-2, this process has become more accessible — but still requires a convenient interface and computational resources that not all researchers have at hand. If you’ve ever wished for a smoother workflow to get AlphaFold models from sequences, especially when you’re dealing with monomers or multimers, this post is for you.
Why AlphaFold-2 in SAMSON?
SAMSON integrates the popular AlphaFold-2 structure prediction engine with an easy-to-use graphical interface. Instead of local installations and hardware concerns, SAMSON connects you directly to the cloud, where you can submit AlphaFold-2 jobs and retrieve results seamlessly. The tool handles both monomer and multimer predictions, lets you choose alignment databases, and returns colored models based on pLDDT scores as soon as they’re done.
How the Workflow Looks
- Go to Home > Predict within the SAMSON platform.
- Select the AlphaFold-2 prediction service.
- Upload your FASTA file(s). You can add one or more sequences.
- Choose the AlphaFold model type (e.g., monomer, multimer) and select the alignment database you prefer.
- Click Start prediction.
Your job gets queued on powerful cloud instances (including A100 GPUs), with secure data transmission throughout.
Understanding the Output
Once the prediction is complete, you’ll see a 3D model directly inside SAMSON. Structures are automatically colorized based on pLDDT values (a confidence metric ranging from 0 to 100). Blue regions signal high confidence, while orange/red vibes might mean areas worth caution or further experimental validation.
You can monitor your jobs in the Interface > Cloud jobs panel or via your online SAMSON Connect account at Account > Jobs.
Cloud-Based Predictions: Cost Strategy
Predictions incur a cost, measured in computing credits, based on selected resources. For academic users, or anyone needing testing credits, SAMSON offers options to get in touch, or you can buy computing credits directly online. Choosing faster hardware may cost more, but can greatly reduce waiting time.
When to Cite AlphaFold
If your publication includes models generated via AlphaFold-2 in SAMSON, don’t forget to cite the original work:
- Jumper et al., Nature. 2021
- And if using multimers: cite the AlphaFold-Multimer paper as well.
Who Is This For?
This workflow is ideal for:
- Structural biologists working on unknown protein sequences
- Researchers integrating modeling into bioinformatics pipelines
- Students and educators looking for intuitive yet powerful prediction tools
- Collaborators needing quick visualization of predicted structures
If you’re tired of navigating command lines or configuring compute clusters, SAMSON’s AlphaFold-2 interface might provide a welcome shortcut.
📚 Learn More
For full step-by-step instructions, additional formats, and related tools, check out the official Biomolecular Structure Prediction documentation page.
SAMSON and all SAMSON Extensions are free for non-commercial use. You can download SAMSON here.