Balancing multiple objectives in conformation sampling to control decoy diversity in template-free protein structure prediction

Abstract

Background Computational approaches for the determination of biologically-active/native three-dimensional structures of proteins with novel sequences have to handle several challenges. The (conformation) space of possible three-dimensional spatial arrangements of the chain of amino acids that constitute a protein molecule is vast and high-dimensional. Exploration of the conformation spaces is performed in a sampling-based manner and is biased by the internal energy that sums atomic interactions. Even state-of-the-art energy functions that quantify such interactions are inherently inaccurate and associate with protein conformation spaces overly rugged energy surfaces riddled with artifact local minima. The response to these challenges in template-free protein structure prediction is to generate large numbers of low-energy conformations (also referred to as decoys) as a way of increasing the likelihood of having a diverse decoy dataset that covers a sufficient number of local minima possibly housing near-native conformations. Results In this paper we pursue a complementary approach and propose to directly control the diversity of generated decoys. Inspired by hard optimization problems in high-dimensional and non-linear variable spaces, we propose that conformation sampling for decoy generation is more naturally framed as a multi-objective optimization problem. We demonstrate that mechanisms inherent to evolutionary search techniques facilitate such framing and allow balancing multiple objectives in protein conformation sampling. We showcase here an operationalization of this idea via a novel evolutionary algorithm that has high exploration capability and is also able to access lower-energy regions of the energy landscape of a given protein with similar or better proximity to the known native structure than several state-of-the-art decoy generation algorithms. Conclusions The presented results constitute a promising research direction in improving decoy generation for template-free protein structure prediction with regards to balancing of multiple conflicting objectives under an optimization framework. Future work will consider additional optimization objectives and variants of improvement and selection operators to apportion a fixed computational budget. Of particular interest are directions of research that attenuate dependence on protein energy models.

Decoy Ensemble Reduction in Template-free Protein Structure Prediction

“…Evaluation is carried out on two datasets. The first is a benchmark dataset of 10 proteins of varying lengths and folds that are used widely for evaluation [17,[25][26][27]. The second contains 10 hard, free-modeling target domains from CASP12 and CASP13.…”

“…Good advances have been made in decoy generation [10,[16][17][18][23][24][25][26]. These advances are documented in the Critical Assessment of protein Structure Prediction (CASP), which is a biennial community experiment/competition that assesses progress in PSP in several categories, including the template-free/free modeling category [9].…”

“…Great progress in software and hardware has made it less costly to generate decoys. Algorithms operating under the umbrella of evolutionary computation can generate hundreds of thousands of decoys [17,[24][25][26]. Decoy selection algorithms tasked with analyzing decoys now may have to additionally deal with a data size issue.…”

Learning Reduced Latent Representations of Protein Structure Data

“…We employ 19 proteins of varying lengths (53 to 146 amino acids long) and folds (α, β, α + β, and coil). These proteins are typically employed to evaluate debuting decoy generation algorithms for template-free PSP [36,37]. On each protein, we run the Rosetta AbInitio protocol in an embarrassingly parallel manner to obtain an ensemble of 50, 000 − 60, 000 decoys per protein.…”

Deep Ranking in Template-free Protein Structure Prediction

“…The space of possible tertiary structures of a given amino-acid sequence is vast and high-dimensional; protein molecules are inherently plastic and undergo both continuous and discrete motions in three-dimensional space [47]. Significant advances have been made in this regard and have resulted in many software packages and methods, such as AlphaFold [46], Rosetta [27], Quark [53], and many others [37][38][39]54]. A key common characteristic of these approaches is the discretization of the search space by relying on the concept of structural fragments distilled from experimentally-determined native structures and utilized to assemble novel structures of a given target.…”