Predicting protein structures is a recurring challenge for many molecular modelers, often involving complex workflows, heavy computation, and multiple configurations. AlphaFold-2 has revolutionized the field by improving the accuracy of protein structure predictions, but its integration in everyday modeling workflows can still feel distant — especially if you’re working with existing software pipelines or lack local computing power.
Thankfully, SAMSON offers a built-in integration for AlphaFold-2 through its Biomolecular Structure Prediction extension. If you haven’t tried this yet, here’s a closer look at how this module streamlines the process and lets you generate reliable structure predictions in just a few steps — without leaving the SAMSON environment.
What Problem Does This Solve?
You no longer need to:
- Set up AlphaFold-2 manually or through command-line interfaces
- Worry about dependencies or model versions
- Deploy large cloud instances yourself
If you’ve ever spent hours trying to ensure compatible software environments for AlphaFold (CUDA versions, RAM availability, Python packages, etc.), this solution offers an immediate alternative.
How It Works in SAMSON
To use AlphaFold-2 in SAMSON:
- Open Home > Predict.
- Select the AlphaFold-2 option.
- Load one or more FASTA files.
- Choose the prediction model (monomer, multimer, etc.).
- Choose the multiple sequence alignment (MSA) database.
- Click Start prediction.
Predictions are carried out in the cloud, with results automatically displayed in SAMSON when they’re ready. For more transparency, you can track them in Interface > Cloud jobs or directly via your SAMSON Connect account.
Visual Feedback & pLDDT Scores
Once your prediction is loaded into SAMSON, structures are colorized based on their pLDDT scores — offering immediate visual insight into the confidence of the predicted atomic positions. pLDDT helps modelers quickly assess which regions are modeled with certainty and which require caution or further validation.
Choices of Compute Performance
You have the option to select computing power for your job. SAMSON supports powerful GPU instances, including A100s, giving you flexibility in terms of both cost and performance. This is especially helpful when predicting larger or more complex sequences. Computing credits are required, but you can purchase or request them easily through the SAMSON platform.
Shared Benefits
This integration doesn’t just simplify the prediction process — it also integrates your results natively into the SAMSON environment. You can immediately analyze, visualize, or simulate your predicted structures, which means fewer file exports and tool switching. For collaborative work, keeping everything in one environment minimizes miscommunications and versioning issues.
Finally, when using the AlphaFold-2 service in your research, remember to cite the original AlphaFold paper: Jumper et al. (2021).
Who Is This For?
This workflow is valuable for:
- Biotech or pharma teams needing quick structure predictions
- Students or early-career researchers without access to HPCs
- Educators introducing structural biology in practical courses
To learn more, 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.