Despite considerable progress, ab initio protein structure prediction remains suboptimal. A crowdsourcing approach is the online puzzle video game Foldit, that provided several useful results that matched or even outperformed algorithmically computed solutions. Using Foldit, the WeFold crowd had several successful participations in the Critical Assessment of Techniques for Protein Structure Prediction. Based on the recent Foldit standalone version, we trained a deep reinforcement neural network called DeepFoldit to improve the score assigned to an unfolded protein, using the Q-learning method with experience replay. This paper is focused on model improvement through hyperparameter tuning. We examined various implementations by examining different model architectures and changing hyperparameter values to improve the accuracy of the model. The new model hyper-parameters also improved its ability to generalize. Initial results, from the latest implementation, show that given a set of small unfolded training proteins, DeepFoldit learns action sequences that improve the score both on the training set and on novel test proteins. Our approach combines the intuitive user interface of Foldit with the efficiency of deep reinforcement learning.
De novo protein structure prediction from amino acid sequence is one of the most challenging problems in computational biology. As one of the extensively explored mathematical models for protein folding, Hydrophobic-Polar (HP) model enables thorough investigation of protein structure formation and evolution. Although HP model discretizes the conformational space and simplifies the folding energy function, it has been proven to be an NP-complete problem. In this paper, we propose a novel protein folding framework FoldingZero, self-folding a de novo protein 2D HP structure from scratch based on …
Ben-Gurion University of the Negev（本-古里安大学）；The University of British Columbia（不列颠哥伦比亚大学）
Recent advancements in machine learning techniques for protein folding motivate better results in its inverse problem -- protein design. In this work we introduce a new graph mimetic neural network, MimNet, and show that it is possible to build a reversible architecture that solves the structure and design problems in tandem, allowing to improve protein design when the structure is better estimated. We use the ProteinNet data set and show that the state of the art results in protein design can be improved, given recent architectures for protein folding.
Cotranslational folding depends on the folding speed and stability of the nascent protein. It remains difficult, however, to predict which proteins cotranslationally fold. Here, we simulate evolution of model proteins to investigate how native structure influences evolution of cotranslational folding. We developed a model that connects protein folding during and after translation to cellular fitness. Model proteins evolved improved folding speed and stability, with proteins adopting one of two strategies for folding quickly. Low contact order proteins evolve to fold cotranslationally. Such pro…
Intrinsically disordered proteins participate in many biological processes by folding upon binding with other proteins. However, coupled folding and binding processes are not well understood from an atomistic point of view. One of the main questions is whether folding occurs prior to or after binding. Here we use a novel unbiased high-throughput adaptive sampling approach to reconstruct the binding and folding between the disordered transactivation domain of \mbox{c-Myb} and the KIX domain of the CREB-binding protein. The reconstructed long-term dynamical process highlights the binding of a sh…
Exploring and understanding the protein-folding problem has been a long-standing challenge in molecular biology. Here, using molecular dynamics simulation, we reveal how parallel distributed adjacent planar peptide groups of unfolded proteins fold reproducibly following explicit physical folding codes in aqueous environments due to electrostatic attractions. Superfast folding of protein is found to be powered by the contribution of the formation of hydrogen bonds. Temperature-induced torsional waves propagating along unfolded proteins break the parallel distributed state of specific amino acid…
We investigated the impact of hydrodynamic interactions (HI) on protein folding using a coarse-grained model. The extent of the impact of hydrodynamic interactions, whether it accelerates, retards, or has no effect on protein folding, has been controversial. Together with a theoretical framework of the energy landscape theory (ELT) for protein folding that describes the dynamics of the collective motion with a single reaction coordinate across a folding barrier, we compared the kinetic effects of HI on the folding rates of two protein models that use a chain of single beads with distinctive to…
De novo prediction of protein folding is an open scientific challenge. Many folding models and force fields have been developed, yet all face difficulties converging to native conformations. Hydrophobicity scales (HSs) play a crucial role in such simulations as they define the energetic interactions between protein residues, thus determining the energetically favorable conformation. While many HSs have been developed over the years using various methods, it is surprising that the scales show very weak consensus in their assignment of hydrophobicity indexes to the various residues. In this work…
Predicting the three-dimensional (3D) functional structures of proteins remains an important computational milestone in molecular biology to be achieved. This feat is hinged on a clear understanding of the mechanism which proteins use to fold into their native structures. Since Levinthal's paradox, there has been a lot of progress in understanding this mechanism. Most of the earlier attempts were caught between assigning either hydrophobic interactions or hydrogen bonding as the dominant folding force. However, a consensus now seems to be emerging about hydrogen bonding being a stronger force.…
Protein folding is a phenomenon that has been studied for about 50 years and still remains as an unsolved problem. The main feature of this process is that it occurs as an all or none process, so a protein, can jump directly between folded and unfolded states. It is proposed in this manuscript that the cooperative phenomena associated to protein folding has its origin in the synchronization oscillating phases of the instantaneously induced dipoles that gives rise to the van der Waals London dispersion interaction. When the oscillation of the induced dipoles is synchronized an enhanced interact…
Recent advances in distance-based protein folding have led to a paradigm shift in protein structure prediction. Through sufficiently precise estimation of the inter-residue distance matrix for a protein sequence, it is now feasible to predict the correct folds for new proteins much more accurately than ever before. Despite the exciting progress, a dedicated visualization system that can dynamically capture the distance-based folding process is still lacking. Most molecular visualizers typically provide only a static view of a folded protein conformation, but do not capture the folding process.…
The intricate three-dimensional geometries of protein tertiary structures underlie protein function and emerge through a folding process from one-dimensional chains of amino acids. The exact spatial sequence and configuration of amino acids, the biochemical environment and the temporal sequence of distinct interactions yield a complex folding process that cannot yet be easily tracked for all proteins. To gain qualitative insights into the fundamental mechanisms behind the folding dynamics and generic features of the folded structure, we propose a simple model of structure formation that takes …
A complex network approach to protein folding is proposed. The graph object is the network of shortcut edges present in a native-state protein (SCN0). Although SCN0s are found via an intuitive message passing algorithm (S. Milgram, Psychology Today, May 1967 pp. 61-67), they are meaningful enough that the logarithm form of their contact order (SCN0_lnCO) correlates significantly with protein kinetic rates, regardless of protein size. Further, the clustering coefficient of a SCN0 (CSCN0) can be used to combine protein segments iteratively within the Restricted Binary Collision model to form the…
Contact-assisted protein folding has made very good progress, but two challenges remain. One is accurate contact prediction for proteins lack of many sequence homologs and the other is that time-consuming folding simulation is often needed to predict good 3D models from predicted contacts. We show that protein distance matrix can be predicted well by deep learning and then directly used to construct 3D models without folding simulation at all. Using distance geometry to construct 3D models from our predicted distance matrices, we successfully folded 21 of the 37 CASP12 hard targets with a medi…
Protein folding is an indispensable process for the majority of proteins after their synthesis from ribosomes in the cell. Most in vitro protein folding studies have focused on single-domain proteins. Hence, it is important to understand the folding process of multi-domain proteins, especially when domains are discontinuous. We choose the Maltose binding protein (MBP) as a model system. In particular, we studied a mutant of MBP that folds slowly. Here, using two- and three-color single-molecule Foerster resonance energy transfer (smFRET) experiments, we study the refolding of both the domains …
