I am looking at B-factors in two different PDB protein structures and I wonder if these B-factors are suitable for direct inter-PDBs comparisons, say to tell whether a particular loop in protein A is more flexible than another loop in protein B . Thanks in advance for your help.
Hi Pedro,
Could you please specify which two structures are you comparing? I mean, are they two unrelated proteins (low structural homology), or are they different versions of the same protein (for example, an holoprotein, and its complexed counterpart)?
B-factors measure atom mobility inside a 3-D dynamic structure. They are related to atomic displacement, but also to resolution, so an atom's B-factor may be high not only because of high mobility, but also because of low structure resolution (if atom position was wrongly placed when inferring structure). Unless you have a high resolution structure, you usually don't know if your atom's B-factor is high due to high mobility, or due to poor resolution. That is why, if you have structures with different resolutions, I definitely unrecommend you comparing their B-factors. B-factor comparisons may be very tricky, so you should be very careful when interpreting the results.
But I could think of a situation in which B-factor comparison may be valuable: when comparing different states from the same protein (complexed-uncomplexed, for example). Even in this case, I would normalize the B-factors prior to comparison. In this case, you could identify protein regions whose mobility is reduced/increased because of structural changes, like when a holoprotein binds to a ligand.
I hope this helps.
Hi Pedro,
Could you please specify which two structures are you comparing? I mean, are they two unrelated proteins (low structural homology), or are they different versions of the same protein (for example, an holoprotein, and its complexed counterpart)?
B-factors measure atom mobility inside a 3-D dynamic structure. They are related to atomic displacement, but also to resolution, so an atom's B-factor may be high not only because of high mobility, but also because of low structure resolution (if atom position was wrongly placed when inferring structure). Unless you have a high resolution structure, you usually don't know if your atom's B-factor is high due to high mobility, or due to poor resolution. That is why, if you have structures with different resolutions, I definitely unrecommend you comparing their B-factors. B-factor comparisons may be very tricky, so you should be very careful when interpreting the results.
But I could think of a situation in which B-factor comparison may be valuable: when comparing different states from the same protein (complexed-uncomplexed, for example). Even in this case, I would normalize the B-factors prior to comparison. In this case, you could identify protein regions whose mobility is reduced/increased because of structural changes, like when a holoprotein binds to a ligand.
I hope this helps.
Juan, thanks for your reply. I want to compare B-factors of unrelated proteins.
A comparison of B-factors for different proteins solved at different resolutions using different refinement strategies would be meaningless. As juan says, comparison of the same protein might be doable, but only after some normalisation and probably only in a qualitative manner (eg , this loop has higher than average B's in structure A, but lower than average in structure B)
In some structures (pre-2011), the B factors may be residual rather than full, so the data in the B-factor column can even represent different things. Probable residual Bs are flagged under the mmCIF tag _refine.pdbx_TLS_residual_ADP_flag as 'LIKELY RESIDUAL' (or if you are still using PDB format, as REMARK 3 B VALUE TYPE : LIKELY RESIDUAL)
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Proteins