Predicting the three-dimensional structure of proteins is a common challenge in molecular modeling. Even with recent AI-based methods like AlphaFold-2, understanding the quality of predictions across a structure remains critical for interpreting results or deciding which regions of a protein to explore further. If you’ve ever wondered which parts of a predicted structure you can trust the most—good news: this is exactly what SAMSON can help you with.
In this post, we’ll look at how SAMSON, through the Biomolecular Structure Prediction extension, allows you to not only predict protein structures using AlphaFold-2 but also visualize confidence levels in your predictions using pLDDT scores.
What is pLDDT, and why does it matter?
pLDDT stands for “predicted Local Distance Difference Test”. It is a per-residue confidence metric ranging from 0 to 100 that indicates how reliable AlphaFold-2 believes its prediction is for each part of the protein. Higher scores mean greater confidence. In practice, structures with pLDDT scores above 90 are considered highly accurate, while regions with lower scores should be treated with caution.
Visualizing Confidence with SAMSON
When you perform a structure prediction using AlphaFold-2 in SAMSON, the software goes a step further: it automatically colorizes the predicted protein structure based on the pLDDT values—if the information is available in the prediction output.
This visualization allows you to immediately see which regions of your protein are predicted with high confidence and which are uncertain. It’s a small step that can make a big difference, especially for modelers planning mutagenesis or docking simulations.
Once the prediction is complete…
- Navigate to Interface > Cloud Jobs in SAMSON to view your submitted jobs.
- Open your structure directly in SAMSON.
- If the file contains pLDDT data, the residues will be automatically color-coded: blue for high confidence, orange/red for lower confidence.
This saves you a manual step (no need to map scores manually), and it helps you focus your efforts where the model is strongest.
Try It Out With Your Own FASTA Sequences
To run an AlphaFold-2 structure prediction in SAMSON:
- Go to Home > Predict.
- Choose AlphaFold-2 as your service.
- Upload one or more FASTA files.
- Select the modeling type (e.g., monomer, multimer) and a multiple sequence alignment (MSA) database.
- Click Start prediction.
Your model will be computed in the Cloud using GPUs like the A100. You need computing credits – reach out to us at contact@samson-connect.net or purchase them via the credit portal.
Conclusion
Having access to a model is useful. Knowing how much of that model you can trust is essential. SAMSON takes confidence visualization one step further by simplifying how pLDDT values are displayed, allowing you to identify stable regions at a glance and make informed decisions.
Whether you’re assessing structural models for drug design, enzyme engineering, or basic research, this color-coded view can serve as a valuable guide.
📖 Learn more from the full documentation page.
SAMSON and all SAMSON Extensions are free for non-commercial use. You can get SAMSON at https://www.samson-connect.net.