Evaluation of coarse grained models for hydrogen bonds in proteins

Abstract

Backbone hydrogen bonds contribute very importantly to the stability of proteins and therefore they must be appropriately represented in protein folding simulations. Simple models are frequently used in theoretical approaches to this process, but their simplifications are often confronted with the need to be true to the physics of the interactions. Here we study the effects of different levels of coarse graining in the modeling of backbone hydrogen bonds. We study three different models taken from the bibliography in a twofold fashion. First, we calculate the hydrogen bonds in 2gb1, an (alpha + beta)-protein, and see how different backbone representations and potentials can mimic the effects of real hydrogen bonds both in helices and sheets. Second, we use an evolutionary method for protein fragment assembly to locate the global energy minimum for a set of small beta-proteins with these models. This way, we assess the effects of coarse graining in hydrogen bonding models and show what can be expected from them when used in simulation experiments.

Energy minimizations with a combination of two knowledge‐based potentials for protein folding

“…This restrictions are related to the geometry of hydrogen‐bonded structures in actual proteins. As we thoroughly explained in our previous work with this potential,27 these restrictions severely constrain the formation of these interactions in the model, thus emulating the directional nature of the hydrogen bond. After the conditions are fulfilled for residues i and j , the energy is calculated as a sum of two terms 28.…”

“…The method that we use here is the same that we employed in previous studies 14, 27, 31. Therefore, we just briefly describe it for the reader's convenience.…”

“…To find the hydrogen bonding potential we also assessed three models with different level of coarse graining 27. We found that this contribution is specially well‐crafted for simulation experiments in the CABS model,28 as developed by Kolinski.…”

A multiscale approach to model hydrogen bonding: The case of polyamide

“…[18][19][20] The atoms involved in these highly directional non-bonded interactions are grouped into single (or different) CG beads and the interaction is averaged away with other non-bonded interactions. In the case of biopolymers where the hydrogen bond network is responsible for their three dimensional structure, this problem has been solved developing specific CG force field parameters for each aminoacid or DNA base; [21][22][23] however, such highly targeted approaches cannot be used for other synthetic materials whose 3D structure is not known a priori. The formation and disruption of the hydrogen bonding network are also intrinsically related to the system dynamics which can be altered when the model lacks an explicit treatment of it.…”

A refined hydrogen bond potential for flexible protein models

“…Consequently, a wide range of intermediate resolution models has been proposed recently. 10,[16][17][18][19] Combined with other potentials for the rest of energetic interactions or applied to rigid fragments, 20 they give good results in most cases, leading to recognizable structures that are sensitive to the system conditions.…”