Molecular modelers often face a common challenge: how to reliably predict protein structures with tools that are accessible, secure, and flexible enough to support a variety of research workflows. With the growing popularity of AlphaFold-2, the ability to make structure predictions from sequence data using a reliable interface is now in higher demand than ever.
In this blog post, we take a closer look at how SAMSON, the integrative molecular design platform, integrates AlphaFold-2 predictions into a visual, cloud-connected interface, allowing you to move seamlessly from sequence to structure — all from your desktop.
What Problem Are We Solving?
Many researchers need to predict protein structures quickly and use the results within an integrated modeling environment. However, relying on web services alone often makes managing inputs and outputs less transparent, and integrating results with other data can be tedious. SAMSON provides an end-to-end visual platform where you can:
- Input sequences via FASTA files
- Select prediction models (e.g. monomer, multimer)
- Choose the multiple sequence alignment (MSA) database
- Review and visualize predictions in one place
How to Use AlphaFold-2 in SAMSON
To get started, make sure you have installed the Biomolecular Structure Prediction extension in SAMSON. Then follow these steps:
- In SAMSON, open the Home > Predict panel.
- Select AlphaFold-2 as your prediction service.
- Add one or several FASTA files with your sequences.
- Choose the appropriate AlphaFold model (e.g. monomer, multimer) and the MSA database to use for alignment.
- Click Start prediction.
The task is then sent to the cloud using a secure connection. You can monitor the status in the Interface > Cloud jobs section in SAMSON, or through the SAMSON Connect Account Jobs page.
Cloud GPUs and Cost
AlphaFold computations are resource-intensive. In SAMSON, predictions can be performed on powerful A100 GPUs in the cloud, offering fast turnaround times. Prediction jobs require computing credits; you can purchase them or request academic credits depending on your situation.
After the Prediction
Once the structure is returned, SAMSON colorizes it based on pLDDT scores — when available — making it easier to assess prediction confidence visually. You can immediately inspect, edit, or use the predicted structures in simulations from within the platform.
What About Publishing?
If you plan to publish results derived from AlphaFold predictions, be sure to cite the original AlphaFold paper. For multimers, cite the corresponding AlphaFold-Multimer publication mentioned in the documentation.
By integrating AlphaFold-2 directly into the modeling workflow, SAMSON makes it easier for researchers to go from amino acid sequence to 3D validated models — while retaining data control throughout the process.
To learn more about prediction services in SAMSON, including options beyond AlphaFold-2 like Boltz-2 and Chai-1, visit the official documentation page: 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.