RSMD in homology modelling protein

Question

[l.souza]

Hello everyone,

I am modeling a protein through homology approach. My protein has a large disordered structure (~46% of coil). After the refinement, the RSMD between the template and the final model was 3.68Å. Is it too high?

Thanks in advance!

Answer 1

[Joe]

3.6 seems reasonably good, but visually inspect the match to make sure. It’s possible to get low RMSD values even if the proteins don’t super impose well.

You may want to look at another metric using a program such as TM-align, since there are well documented issues with RMSDs as a lone measurement of quality.

[cpad0112]

To my knowledge, RMSD < 1.5 seems to be good for proteins of same family (for eg multi copper oxidase family). Some others consider RMSD <2 is good enough. However, i guess this depends up on the evolutionary conservation within the group.

[Joe]

It also depends on what you intend to do with the model. If the higher RMSD is contributed to from mostly disordered loops, but you’re interested in a well conserved active site, you models might be good enough in a localised region to be useful, if not overall.

[cpad0112]

That is true. Local RMSD vs Global RMSD.

[l.souza]

The TM-score is 0.64319. So, according to the reference value, the protein is in about the same fold, right? Visually, they seem to be well imposed, with isolated places not so good (most of them are presented as coil).

[Joe]

If the regions of the protein you’re interested in look OK, you’re probably alright to proceed cautiously - it really depends on what your actual question is though?