Proteins play essential roles in cellular processes, and accurate three-dimensional structures are critical for understanding function and enabling drug discovery. High-resolution methods such as cryo-electron microscopy (cryo-EM) and X-ray crystallography can provide detailed structural information. However, many proteins are characterized only by low-resolution or sparse data, which are difficult to translate into accurate atomic models. Mass spectrometry-based approaches such as ion mobility (IM-MS) provide global structural information through collisional cross section (CCS) but lack atomistic detail. Here, we present CRIM (cryo-EM + IM-MS), an integrative Rosetta scoring function that combines low-resolution cryo-EM density with IM-MS-derived CCS restraints to improve monomeric protein structure prediction. CRIM integrates the REF2015 energy with CCS agreement (via PARCS) and an electron density term (elecdensfast). On an ideal data set of 60 proteins, CRIM reduced the mean RMSD from 3. 65 to 2. 90 Å and increased the TM-score from 0. 88 to 0. 90. On an experimental data set of 54 proteins, CRIM improved model selection, lowering the RMSD from 6. 65 to 4. 38 Å and increasing the TM-score from 0. 73 to 0. 79. Compared to AlphaFold3, CRIM produced competitive predictions and outperformed it for select challenging targets where sparse experimental data provide strong constraints. The CRIM score function is freely available within Rosetta and provides a practical framework for integrating complementary cryo-EM and IM-MS data to improve protein structure prediction.
Howard et al. (Wed,) studied this question.