On the use of low-resolution data to improve structure prediction of proteins and protein complexes

We present a systematic study of the ability of low-resolution experimental data, when combined with physical/statistical scoring functions, to improve the quality of theoretical structural models of proteins and protein complexes. Particularly, we have analyzed in detail the "extra value" added to the theoretical models by: electrospray mass spectrometry (ESIMS), small-angle X-ray scattering (SAXS), and hydrodynamic measurements. We found that any low-resolution structural data, even when (as in the case of mass spectrometry) obtained in conditions far from the physiological ones, help to improve the quality of theoretical models, but not all the coarse-grained experimental results are equally rich in information. The best results are always obtained when using SAXS data as experimental constraints, but either hydrodynamics or gas phase CCS data contribute to improving model prediction. The combination of suitable scoring functions and broadly available low-resolution structural data (technically easier to obtain) yields structural models that are notably close to the real structures.