profile

ProteinReach by Dr. Omar

Crystal structure vs. Boltz2 Prediction

Published about 1 month ago • 1 min read

I used Boltz-2 to predict the structure of a ligand-bound mutant protein complex (AtFAAH S305A bound to OdDHL) which I worked with during my PhD. The ligand, OdDHL, is a quorum sensing molecule. The crystal structure of this complex was previously resolved at 3.5 Å resolution, so not the highest quality.
​
When comparing the predicted and experimental structures using the superimpose tool in PyMOL, I noticed that the Boltz-2 prediction successfully folds certain regions that were unresolved in the crystal structure; which is fantastic!
​
However, I observed some discrepancies in the ligand’s pose:
​
-I found that the Boltz-2 positioned the ligand further away from S305A (a catalytic serine mutated to alanine and used as reference point) compared to its location in the crystal structure.
​
-The ligand at the fatty acid tail drifted in the boltz2 prediction relative to the pose in the crystal structure therefore affecting the types of residues surrounding the pocket.
​
What would you do? Perform MDS perhaps to assess if poses at some point overlap? Which structure do you trust the most?

Thank you,

Best,

Omar.

© ProteinReach|2026 SciLearningWorkshops LLC

​
​Unsubscribe · Preferences​

ProteinReach by Dr. Omar

Your Weekly Protein Modeling Briefing

A weekly newsletter delivering the latest protein modeling tools and strategies to streamline your in silico processes, reduce wet lab workload, and save your organization time, effort, and resources.

Read more from ProteinReach by Dr. Omar

I always use MMGBSA and MMPBSA as initial binding energy methods. Then use free energy perturbation (FEP) method to assess how accurate my two initial measurements were. So, the sequence of usage goes like this: 1) MMGBSA2) MMPBSA3) FEP The only downside of using FEP is that is computationally expensive. And not many groups have the possibility of using that one. But if you do, please go ahead and use it. Thank you, Best, Omar. © ProteinReach|2026 SciLearningWorkshops LLC Unsubscribe ·...

1 day ago • 1 min read

During my PhD, I was lost in GROMACS MDS outputs. If you’re just getting started, that huge number of files can be overwhelming; I know that’s how I felt. To help with that, I put together a quick reference guide covering the most common GROMACS file extensions and what they do. Please save it and share it with a colleague who’s just starting with MD simulations! Thank you, Best, Omar. © ProteinReach|2026 SciLearningWorkshops LLC Unsubscribe · Preferences

about 1 month ago • 1 min read

Am I the only one who sees a huge disconnect between protein designs and how efficiently they can be executed in the wet-lab? What do you think it’s missing?What could be done for it be improved? Are the datasets used to train those models not good enough anymore?What do you think? Best, Omar. © ProteinReach|2026 SciLearningWorkshops LLC Unsubscribe · Preferences

about 2 months ago • 1 min read