Vagabond: bond-based parametrization reduces overfitting for refinement of proteins.
Ontology highlight
ABSTRACT:
SUBMITTER: Ginn HM
PROVIDER: S-EPMC8025884 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Flexible fitting computational algorithms are often useful to interpret low-resolution maps of many macromolecular complexes generated by electron microscopy (EM) imaging. One such atomistic simulation technique is molecular dynamics flexible fitting (MDFF), which has been widely applied to generate structural models of large ribonucleoprotein assemblies such as the ribosome. We have previously shown that MDFF simulations of globular proteins are sensitive to the resolution of the target EM map and the strength of restraints used to preserve the secondary structure elements during fitting (Vashisth, H.; et al. Structure 2012, 20, 1453-1462). In this work, we aim to systematically examine the quality of structural models of various nucleic acids obtained via MDFF by varying the map resolution and the strength of structural restraints. We also demonstrate how an enhanced conformational sampling technique for proteins, temperature-accelerated molecular dynamics (TAMD), can be combined with MDFF for the structural refinement of nucleic acids in EM maps. Finally, we also demonstrate application of TAMD-assisted MDFF (TAMDFF) on a RNA/protein complex and suggest that TAMDFF is a viable strategy for enhanced conformational fitting in target maps of ribonucleoprotein complexes.
| S-EPMC3690198 | biostudies-literature
Project description:Combined quantum mechanical and molecular mechanical (QM/MM) methods are the most powerful available methods for high-level treatments of subsystems of very large systems. The treatment of the QM-MM boundary strongly affects the accuracy of QM/MM calculations. For QM/MM calculations having covalent bonds cut by the QM-MM boundary, it has been proposed previously to use a scheme with system-specific tuned fluorine link atoms. Here, we propose a broadly parametrized scheme where the parameters of the tuned F link atoms depend only on the type of bond being cut. In the proposed new scheme, the F link atom is tuned for systems with a certain type of cut bond at the QM-MM boundary instead of for a specific target system, and the resulting link atoms are call bond-tuned link atoms. In principle, the bond-tuned link atoms can be as convenient as the popular H link atoms, and they are especially well adapted for high-throughput and accurate QM/MM calculations. Here, we present the parameters for several kinds of cut bonds along with a set of validation calculations that confirm that the proposed bond-tuned link-atom scheme can be as accurate as the system-specific tuned F link-atom scheme.
| S-EPMC6100187 | biostudies-literature
Project description:We propose a novel force-field-parametrization procedure that fits the parameters of potential functions in a manner that the pair distribution function (DF) of molecules derived from candidate parameters can reproduce the given target DF. Conventionally, approaches to minimize the difference between the candidate and target DFs employ radial DFs (RDF). RDF itself has been reported to be insufficient for uniquely identifying the parameters of a molecule. To overcome the weakness, we introduce energy DF (EDF) as a target DF, which describes the distribution of the pairwise energy of molecules. We found that the EDF responds more sensitively to a small perturbation in the pairwise potential parameters and provides better fitting accuracy compared to that of RDF. These findings provide valuable insights into a wide range of coarse graining methods, which determine parameters using information obtained from a higher-level calculation than that of the developed force field. © 2019 The Authors. Journal of Computational Chemistry published by Wiley Periodicals, Inc.
| S-EPMC6790663 | biostudies-literature
Project description:The phenomenon of benign overfitting is one of the key mysteries uncovered by deep learning methodology: deep neural networks seem to predict well, even with a perfect fit to noisy training data. Motivated by this phenomenon, we consider when a perfect fit to training data in linear regression is compatible with accurate prediction. We give a characterization of linear regression problems for which the minimum norm interpolating prediction rule has near-optimal prediction accuracy. The characterization is in terms of two notions of the effective rank of the data covariance. It shows that overparameterization is essential for benign overfitting in this setting: the number of directions in parameter space that are unimportant for prediction must significantly exceed the sample size. By studying examples of data covariance properties that this characterization shows are required for benign overfitting, we find an important role for finite-dimensional data: the accuracy of the minimum norm interpolating prediction rule approaches the best possible accuracy for a much narrower range of properties of the data distribution when the data lie in an infinite-dimensional space vs. when the data lie in a finite-dimensional space with dimension that grows faster than the sample size.
| S-EPMC7720150 | biostudies-literature
Project description:Overfitting is one of the critical problems in developing models by machine learning. With machine learning becoming an essential technology in computational biology, we must include training about overfitting in all courses that introduce this technology to students and practitioners. We here propose a hands-on training for overfitting that is suitable for introductory level courses and can be carried out on its own or embedded within any data science course. We use workflow-based design of machine learning pipelines, experimentation-based teaching, and hands-on approach that focuses on concepts rather than underlying mathematics. We here detail the data analysis workflows we use in training and motivate them from the viewpoint of teaching goals. Our proposed approach relies on Orange, an open-source data science toolbox that combines data visualization and machine learning, and that is tailored for education in machine learning and explorative data analysis.
| S-EPMC7932115 | biostudies-literature
Project description:Zervamicin IIB (Zrv-IIB) is a channel-forming peptaibol antibiotic of fungal origin. The measured transhydrogen bond (3h)J(NC') couplings in methanol solution heaving average value of -0.41 Hz indicate that the stability of the Zrv-IIB helix in this milieu is comparable to the stability of helices in globular proteins. The N-terminus of the peptide forms an alpha-helix, whereas 3(10)-helical hydrogen bonds stabilize the C-terminus. However, two weak transhydrogen bond peaks are observed in a long-range HNCO spectrum for HN Aib(12). Energy calculations using the Empirical Conformation Energy Program for Peptides (ECEPP)/2 force field and the implicit solvent model show that the middle of the peptide helix accommodates a bifurcated hydrogen bond that is simultaneously formed between HN Aib(12) and CO Leu(8) and CO Aib(9). Several lowered (3h)J(NC') on a polar face of the helix correlate with the conformational exchange process observed earlier and imply dynamic distortions of a hydrogen bond pattern with the predominant population of a properly folded helical structure. The refined structure of Zrv-IIB on the basis of the observed hydrogen bond pattern has a small ( approximately 20 degrees ) angle of helix bending that is virtually identical to the angle of bending in dodecylphosphocholine (DPC) micelles, indicating the stability of a hinge region in different environments. NMR parameters ((1)HN chemical shifts and transpeptide bond (1)J(NC') couplings) sensitive to hydrogen bonding along with the solvent accessible surface area of carbonyl oxygens indicate a large polar patch on the convex side of the helix formed by three exposed backbone carbonyls of Aib(7), Aib(9), and Hyp(10) and polar side chains of Hyp(10), Gln(11), and Hyp(13). The unique structural features, high helix stability and the enhanced polar patch, set apart Zrv-IIB from other peptaibols (for example, alamethicin) and possibly underlie its biological and physiological properties.
| S-EPMC1304270 | biostudies-literature
Project description:The site frequency spectrum is an important summary statistic in population genetics used for inference on demographic history and selection. However, estimation of the site frequency spectrum from called genotypes introduces bias when working with low-coverage sequencing data. Methods exist for addressing this issue but sometimes suffer from 2 problems. First, they can have very high computational demands, to the point that it may not be possible to run estimation for genome-scale data. Second, existing methods are prone to overfitting, especially for multidimensional site frequency spectrum estimation. In this article, we present a stochastic expectation-maximization algorithm for inferring the site frequency spectrum from NGS data that address these challenges. We show that this algorithm greatly reduces runtime and enables estimation with constant, trivial RAM usage. Furthermore, the algorithm reduces overfitting and thereby improves downstream inference. An implementation is available at github.com/malthesr/winsfs.
| S-EPMC9713400 | biostudies-literature
Project description:Model building and refinement, and the validation of their correctness, are very effective and reliable at local resolutions better than about 2.5 Å for both crystallography and cryo-EM. However, at local resolutions worse than 2.5 Å both the procedures and their validation break down and do not ensure reliably correct models. This is because in the broad density at lower resolution, critical features such as protein backbone carbonyl O atoms are not just less accurate but are not seen at all, and so peptide orientations are frequently wrongly fitted by 90-180°. This puts both backbone and side chains into the wrong local energy minimum, and they are then worsened rather than improved by further refinement into a valid but incorrect rotamer or Ramachandran region. On the positive side, new tools are being developed to locate this type of pernicious error in PDB depositions, such as CaBLAM, EMRinger, Pperp diagnosis of ribose puckers, and peptide flips in PDB-REDO, while interactive modeling in Coot or ISOLDE can help to fix many of them. Another positive trend is that artificial intelligence predictions such as those made by AlphaFold2 contribute additional evidence from large multiple sequence alignments, and in high-confidence parts they provide quite good starting models for loops, termini or whole domains with otherwise ambiguous density.
| S-EPMC10833350 | biostudies-literature
Project description:Quantum-based refinement utilizes chemical restraints derived from quantum-chemical methods instead of the standard parameterized library-based restraints used in refinement packages. The motivation is twofold: firstly, the restraints have the potential to be more accurate, and secondly, the restraints can be more easily applied to new molecules such as drugs or novel cofactors. Here, a new project called Q|R aimed at developing quantum-based refinement of biomacromolecules is under active development by researchers at Shanghai University together with PHENIX developers. The central focus of this long-term project is to develop software that is built on top of open-source components. A development version of Q|R was used to compare quantum-based refinements with standard refinement using a small model system.
| S-EPMC5331472 | biostudies-literature
Project description:The ability to separate analytes with increasingly similar properties drives the field of separation science. One way to achieve such separations is using trapping and streaming dielectrophoresis (DEP), which directly exploits the subtle differences in the electrophysical properties of analytes. The non-uniform fields necessary for DEP can be formed using various insulator shapes in microchannels. Current insulator shapes include triangles, diamonds, circles, and rectangles. However, all of these insulators pose problems for trapping, streaming, and sorting (deflection) as the induced fields/gradients are not behaviorally consistent across the lateral dimension. This leads to analytes experiencing different forces depending on their pathline in the microchannel and result in low resolution separations. Based on an iterative process that explored approximately 40 different insulator shapes, a design was chosen that indicated improved particle streamlines, better trapping efficiency, and consistent electrical environments across the lateral dimension. The design was assessed by simulations where the electric field, gradient of the electric field squared, and the ratio of the two were plotted. The improved design includes a unique new multi-length scale element. The multi-length scale structure streamlines the analyte(s) and improves homogeneity in the lateral dimension, while still achieving high gradients necessary for analyte separation using DEP. The design is calculated to keep analytes on the centerline which should improve resolution, and eliminate extraneous trapping zones. Behaviors consistent with the features of the simulations were observed in proof of principle experiments using representative test probes.
| S-EPMC5507384 | biostudies-literature