Thank you very much Yahia and Sebastian for your great comments.

I see Yahia's point that the structure of the protein in solution may generally be more desirable, but if that were the only consideration most structures would be solved by NMR, which is not the case.

Sebastian, could you please elaborate on why smaller proteins are more amenable for NMR structural calculation? Is it a peak overlap problem, a lack of sufficient restrictions...? And how can related proteins help here (we are not interested in homology modeling)?

Thank you so much for your insights

Sometimes the structure of a protein has been determined by both X-ray crystallography and by NMR, for example cytochrome b5 (PDB-codes 1ehb and 1aqa, respectively). One can then superimpose the structures for example in DeepView (fit -> magic fit) to get an impression how much they differ. In case of helices and strands the difference is very small, loops are less precisely determined. But this one can see even by comparison of the different NMR models.

Thank you all for your insights. If you are reading the discussion and have an opinion, I would love to know. Also, if someone has an economic estimate of solving a typical structure using either approach, I would love to hear.

Thank you very much!

As a rule of thumb, protein structure determination requires about 100 mg of pure protein and 1 year of work for a small team.

Hi Vincent,

Because I don’t know properties of your protein, I recommend you an article I read before, maybe can help you choose the experimental method for determining the structure (https://www.creative-biostructure.com/comparison-of-crystallography-nmr-and-em_6.htm).

Hope it helps!

The 2 methods differ not only with respect to the very different environmental conditions of the experiments(concentrations,co-solvents,temperature,etc) but also in principle.The X-rays (or neutron) scattering method yields a single structure (except for perhaps some ambiguous

regions,like loops,inside it ) also for large ,multidomain proteins,whereas the output of the NMR method is an ensemble of several structures compatible with

NOE signals and established stereochemical constraints.It is practically impossible to know a priori whether the NMR (which is at its best for small, single

domain proteins) yields an average structure close to the x-rays result.Perhaps the best one can do ,is do both plus some computational work in order to understand potential differences and their causes.One advantage of NMR is that one does not need to find appropriate crystallization conditions which can be problematic,and that the protein microenvironment is closer to the natural millieu

Dikeos Mario Soumpasis ,Actually NMR structures of proteins can be determined by magic angle spinning solid state NMR using crystallized protein. There are certain advantages to this technique especially for characterizing membrane proteins. However, NMR structural biologists at my institution prefer not to do this probably because NMR structure determination usually does not require a crystal traditionally.

NMR is considered a more expensive technique because the protein has to be enriched with radionuclides by feeding 15-N labeled Ammonium Chloride and 13-C labeled Glucose to bacteria and this is expensive and the labeling is not always efficient.

For X-ray crystallography, the better your crystal and the better your laser the more solvable your diffraction patterns are. Of course the X-ray source is expensive, but if your institution does not have a high precision X-ray source, form your crystals and get permission to take those crystals to Argonne National Lab to use their Advanced Photon Source if you are in the United States, that is.

Also, it is possible to get more than one state of a protein in a crystal as work in Lorena Beese's lab famously showed a DNA polymerase replicating DNA inside of a crystal.

Adron Ung Thanks for the additional information you have communicated and i would like to have some references to the topics you have mentioned.However I don't see how solid state NMR can discriminate signals from different domains from those within one domain (without knowing the crystal structure) completely unlike X-rays scattering which delivers the whole information( provided one can crystallize the protein) due to interference of signals from the whole crystal and

the availability of methods to determine necessary parameters such as space group,unit cell, number of molecules in the asymmetric cell etc.How do you get all this from magic angles?Concerning expenses I have no idea because I am a theoretician.What I have said pertains to the vast majority of proteins studied.I consider the example from Beese's simply as an exceptional,(albeit very interesting ) finding.