BraggNet: integrating Bragg peaks using neural networks

Revealing the atomic and electronic mechanism of human manganese superoxide dismutase product inhibition

Human manganese superoxide dismutase (MnSOD) is a crucial oxidoreductase that maintains the vitality of mitochondria by converting superoxide (O2●-) to molecular oxygen (O2) and hydrogen peroxide (H2O2) with proton-coupled electron transfers (PCETs). Human MnSOD has evolved to be highly product inhibited to limit the formation of H2O2, a freely diffusible oxidant and signaling molecule. The product-inhibited complex is thought to be composed of a peroxide (O22-) or hydroperoxide (HO2-) species bound to Mn ion and formed from an unknown PCET mechanism. PCET mechanisms of proteins are typically not known due to difficulties in detecting the protonation states of specific residues that coincide with the electronic state of the redox center. To shed light on the mechanism, we combine neutron diffraction and X-ray absorption spectroscopy of the product-bound, trivalent, and divalent states of the enzyme to reveal the positions of all the atoms, including hydrogen, and the electronic configuration of the metal ion. The data identifies the product-inhibited complex, and a PCET mechanism of inhibition is constructed.

The role of Tyr34 in proton-coupled electron transfer of human manganese superoxide dismutase

Human manganese superoxide dismutase (MnSOD) plays a crucial role in controlling levels of reactive oxygen species (ROS) by converting superoxide (O2 ●-) to molecular oxygen (O2) and hydrogen peroxide (H2O2) with proton-coupled electron transfers (PCETs). The reactivity of human MnSOD is determined by the state of a key catalytic residue, Tyr34, that becomes post-translationally inactivated by nitration in various diseases associated with mitochondrial dysfunction. We previously reported that Tyr34 has an unusual pKa due to its proximity to the Mn metal and undergoes cyclic deprotonation and protonation events to promote the electron transfers of MnSOD. To shed light on the role of Tyr34 MnSOD catalysis, we performed neutron diffraction, X-ray spectroscopy, and quantum chemistry calculations of Tyr34Phe MnSOD in various enzymatic states. The data identifies the contributions of Tyr34 in MnSOD activity that support mitochondrial function and presents a thorough characterization of how a single tyrosine modulates PCET catalysis.

The role of Tyr34 in proton-coupled electron transfer of human manganese superoxide dismutase

Human manganese superoxide dismutase (MnSOD) plays a crucial role in controlling levels of reactive oxygen species (ROS) by converting superoxide (O 2 •- ) to molecular oxygen (O 2 ) and hydrogen peroxide (H 2 O 2 ) with proton-coupled electron transfers (PCETs). The reactivity of human MnSOD is determined by the state of a key catalytic residue, Tyr34, that becomes post-translationally inactivated by nitration in various diseases associated with mitochondrial dysfunction. We previously reported that Tyr34 has an unusual pK a due to its proximity to the Mn metal and undergoes cyclic deprotonation and protonation events to promote the electron transfers of MnSOD. To shed light on the role of Tyr34 MnSOD catalysis, we performed neutron diffraction, X-ray spectroscopy, and quantum chemistry calculations of Tyr34Phe MnSOD in various enzymatic states. The data identifies the contributions of Tyr34 in MnSOD activity that support mitochondrial function and presents a thorough characterization of how a single tyrosine modulates PCET catalysis.

The Pixel Anomaly Detection Tool: a user-friendly GUI for classifying detector frames using machine-learning approaches

Data collection at X-ray free electron lasers has particular experimental challenges, such as continuous sample delivery or the use of novel ultrafast high-dynamic-range gain-switching X-ray detectors. This can result in a multitude of data artefacts, which can be detrimental to accurately determining structure-factor amplitudes for serial crystallography or single-particle imaging experiments. Here, a new data-classification tool is reported that offers a variety of machine-learning algorithms to sort data trained either on manual data sorting by the user or by profile fitting the intensity distribution on the detector based on the experiment. This is integrated into an easy-to-use graphical user interface, specifically designed to support the detectors, file formats and software available at most X-ray free electron laser facilities. The highly modular design makes the tool easily expandable to comply with other X-ray sources and detectors, and the supervised learning approach enables even the novice user to sort data containing unwanted artefacts or perform routine data-analysis tasks such as hit finding during an experiment, without needing to write code.

Robust image descriptor for machine learning based data reduction in serial crystallography

Serial crystallography experiments at synchrotron and X-ray free-electron laser (XFEL) sources are producing crystallographic data sets of ever-increasing volume. While these experiments have large data sets and high-frame-rate detectors (around 3520 frames per second), only a small percentage of the data are useful for downstream analysis. Thus, an efficient and real-time data classification pipeline is essential to differentiate reliably between useful and non-useful images, typically known as 'hit' and 'miss', respectively, and keep only hit images on disk for further analysis such as peak finding and indexing. While feature-point extraction is a key component of modern approaches to image classification, existing approaches require computationally expensive patch preprocessing to handle perspective distortion. This paper proposes a pipeline to categorize the data, consisting of a real-time feature extraction algorithm called modified and parallelized FAST (MP-FAST), an image descriptor and a machine learning classifier. For parallelizing the primary operations of the proposed pipeline, central processing units, graphics processing units and field-programmable gate arrays are implemented and their performances compared. Finally, MP-FAST-based image classification is evaluated using a multi-layer perceptron on various data sets, including both synthetic and experimental data. This approach demonstrates superior performance compared with other feature extractors and classifiers.

Revealing the atomic and electronic mechanism of human manganese superoxide dismutase product inhibition

Human manganese superoxide dismutase (MnSOD) is a crucial oxidoreductase that maintains the vitality of mitochondria by converting O 2 ∙ - to O 2 and H 2 O 2 with proton-coupled electron transfers (PCETs). Since changes in mitochondrial H 2 O 2 concentrations are capable of stimulating apoptotic signaling pathways, human MnSOD has evolutionarily gained the ability to be highly inhibited by its own product, H 2 O 2 . A separate set of PCETs is thought to regulate product inhibition, though mechanisms of PCETs are typically unknown due to difficulties in detecting the protonation states of specific residues that coincide with the electronic state of the redox center. To shed light on the underlying mechanism, we combined neutron diffraction and X-ray absorption spectroscopy of the product-bound, trivalent, and divalent states to reveal the all-atom structures and electronic configuration of the metal. The data identifies the product-inhibited complex for the first time and a PCET mechanism of inhibition is constructed.

Revealing the atomic and electronic mechanism of human manganese superoxide dismutase product inhibition

Human manganese superoxide dismutase (MnSOD) is a crucial oxidoreductase that maintains the vitality of mitochondria by converting O 2 ●- to O 2 and H 2 O 2 with proton-coupled electron transfers (PCETs). Since changes in mitochondrial H 2 O 2 concentrations are capable of stimulating apoptotic signaling pathways, human MnSOD has evolutionarily gained the ability to be highly inhibited by its own product, H 2 O 2 . A separate set of PCETs is thought to regulate product inhibition, though mechanisms of PCETs are typically unknown due to difficulties in detecting the protonation states of specific residues that coincide with the electronic state of the redox center. To shed light on the underlying mechanism, we combined neutron diffraction and X-ray absorption spectroscopy of the product-bound, trivalent, and divalent states to reveal the all-atom structures and electronic configuration of the metal. The data identifies the product-inhibited complex for the first time and a PCET mechanism of inhibition is constructed.

Machine learning for analysis of experimental scattering and spectroscopy data in materials chemistry

The rapid growth of materials chemistry data, driven by advancements in large-scale radiation facilities as well as laboratory instruments, has outpaced conventional data analysis and modelling methods, which can require enormous manual effort. To address this bottleneck, we investigate the application of supervised and unsupervised machine learning (ML) techniques for scattering and spectroscopy data analysis in materials chemistry research. Our perspective focuses on ML applications in powder diffraction (PD), pair distribution function (PDF), small-angle scattering (SAS), inelastic neutron scattering (INS), and X-ray absorption spectroscopy (XAS) data, but the lessons that we learn are generally applicable across materials chemistry. We review the ability of ML to identify physical and structural models and extract information efficiently and accurately from experimental data. Furthermore, we discuss the challenges associated with supervised ML and highlight how unsupervised ML can mitigate these limitations, thus enhancing experimental materials chemistry data analysis. Our perspective emphasises the transformative potential of ML in materials chemistry characterisation and identifies promising directions for future applications. The perspective aims to guide newcomers to ML-based experimental data analysis.

Explainable machine learning for diffraction patterns

Serial crystallography experiments at X-ray free-electron laser facilities produce massive amounts of data but only a fraction of these data are useful for downstream analysis. Thus, it is essential to differentiate between acceptable and unacceptable data, generally known as 'hit' and 'miss', respectively. Image classification methods from artificial intelligence, or more specifically convolutional neural networks (CNNs), classify the data into hit and miss categories in order to achieve data reduction. The quantitative performance established in previous work indicates that CNNs successfully classify serial crystallography data into desired categories [Ke, Brewster, Yu, Ushizima, Yang & Sauter (2018). J. Synchrotron Rad.25, 655-670], but no qualitative evidence on the internal workings of these networks has been provided. For example, there are no visualization methods that highlight the features contributing to a specific prediction while classifying data in serial crystallography experiments. Therefore, existing deep learning methods, including CNNs classifying serial crystallography data, are like a 'black box'. To this end, presented here is a qualitative study to unpack the internal workings of CNNs with the aim of visualizing information in the fundamental blocks of a standard network with serial crystallography data. The region(s) or part(s) of an image that mostly contribute to a hit or miss prediction are visualized.

Perfect Crystals: microgravity capillary counterdiffusion crystallization of human manganese superoxide dismutase for neutron crystallography

The NASA mission Perfect Crystals used the microgravity environment on the International Space Station (ISS) to grow crystals of human manganese superoxide dismutase (MnSOD)-an oxidoreductase critical for mitochondrial vitality and human health. The mission's overarching aim is to perform neutron protein crystallography (NPC) on MnSOD to directly visualize proton positions and derive a chemical understanding of the concerted proton electron transfers performed by the enzyme. Large crystals that are perfect enough to diffract neutrons to sufficient resolution are essential for NPC. This combination, large and perfect, is hard to achieve on Earth due to gravity-induced convective mixing. Capillary counterdiffusion methods were developed that provided a gradient of conditions for crystal growth along with a built-in time delay that prevented premature crystallization before stowage on the ISS. Here, we report a highly successful and versatile crystallization system to grow a plethora of crystals for high-resolution NPC.

CNN-Based Laue Spot Morphology Predictor for Reliable Crystallographic Descriptor Estimation

Laue microdiffraction is an X-ray diffraction technique that allows for the non-destructive acquisition of spatial maps of crystallographic orientation and the strain state of (poly)crystalline specimens. To do so, diffraction patterns, consisting of thousands of Laue spots, are collected and analyzed at each location of the spatial maps. Each spot of these so-called Laue patterns has to be accurately characterized with respect to its position, size and shape for subsequent analyses including indexing and strain analysis. In the present paper, several approaches for estimating these descriptors that have been proposed in the literature, such as methods based on image moments or function fitting, are reviewed. However, with the increasing size and quantity of Laue image data measured at synchrotron sources, some datasets become unfeasible in terms of computational requirements. Moreover, for irregular Laue spots resulting, e.g., from overlaps and extended crystal defects, the exact shape and, more importantly, the position are ill-defined. To tackle these shortcomings, a procedure using convolutional neural networks is presented, allowing for a significant acceleration of the characterization of Laue spots, while simultaneously estimating the quality of a Laue spot for further analyses. When tested on unseen Laue spots, this approach led to an acceleration of 77 times using a GPU while maintaining high levels of accuracy.

Machine learning for scattering data: strategies, perspectives and applications to surface scattering

Machine learning (ML) has received enormous attention in science and beyond. Discussed here are the status, opportunities, challenges and limitations of ML as applied to X-ray and neutron scattering techniques, with an emphasis on surface scattering. Typical strategies are outlined, as well as possible pitfalls. Applications to reflectometry and grazing-incidence scattering are critically discussed. Comment is also given on the availability of training and test data for ML applications, such as neural networks, and a large reflectivity data set is provided as reference data for the community.

Cryo neutron crystallography demonstrates influence of RNA 2'-OH orientation on conformation, sugar pucker and water structure

The ribose 2′-hydroxyl is the key chemical difference between RNA and DNA and primary source of their divergent structural and functional characteristics. Macromolecular X-ray diffraction experiments typically do not reveal the positions of hydrogen atoms. Thus, standard crystallography cannot determine 2′-OH orientation (H2′-C2′-O2′-HO2′ torsion angle) and its potential roles in sculpting the RNA backbone and the expansive fold space. Here, we report the first neutron crystal structure of an RNA, the Escherichia coli rRNA Sarcin-Ricin Loop (SRL). 2′-OD orientations were established for all 27 residues and revealed O-D bonds pointing toward backbone (O3′, 13 observations), nucleobase (11) or sugar (3). Most riboses in the SRL stem region show a 2′-OD backbone-orientation. GAGA-tetraloop riboses display a 2′-OD base-orientation. An atypical C2′-endo sugar pucker is strictly correlated with a 2′-OD sugar-orientation. Neutrons reveal the strong preference of the 2′-OH to donate in H-bonds and that 2′-OH orientation affects both backbone geometry and ribose pucker. We discuss 2′-OH and water molecule orientations in the SRL neutron structure and compare with results from a solution phase 10 μs MD simulation. We demonstrate that joint cryo-neutron/X-ray crystallography offers an all-in-one approach to determine the complete structural properties of RNA, i.e. geometry, conformation, protonation state and hydration structure.

Applications of Artificial Intelligence and Machine Learning Algorithms to Crystallization

Artificial intelligence and specifically machine learning applications are nowadays used in a variety of scientific applications and cutting-edge technologies, where they have a transformative impact. Such an assembly of statistical and linear algebra methods making use of large data sets is becoming more and more integrated into chemistry and crystallization research workflows. This review aims to present, for the first time, a holistic overview of machine learning and cheminformatics applications as a novel, powerful means to accelerate the discovery of new crystal structures, predict key properties of organic crystalline materials, simulate, understand, and control the dynamics of complex crystallization process systems, as well as contribute to high throughput automation of chemical process development involving crystalline materials. We critically review the advances in these new, rapidly emerging research areas, raising awareness in issues such as the bridging of machine learning models with first-principles mechanistic models, data set size, structure, and quality, as well as the selection of appropriate descriptors. At the same time, we propose future research at the interface of applied mathematics, chemistry, and crystallography. Overall, this review aims to increase the adoption of such methods and tools by chemists and scientists across industry and academia.

Water structure around a left-handed Z-DNA fragment analyzed by cryo neutron crystallography

Even in high-quality X-ray crystal structures of oligonucleotides determined at a resolution of 1 Å or higher, the orientations of first-shell water molecules remain unclear. We used cryo neutron crystallography to gain insight into the H-bonding patterns of water molecules around the left-handed Z-DNA duplex [d(CGCGCG)]₂. The neutron density visualized at 1.5 Å resolution for the first time allows us to pinpoint the orientations of most of the water molecules directly contacting the DNA and of many second-shell waters. In particular, H-bond acceptor and donor patterns for water participating in prominent hydration motifs inside the minor groove, on the convex surface or bridging nucleobase and phosphate oxygen atoms are finally revealed. Several water molecules display entirely unexpected orientations. For example, a water molecule located at H-bonding distance from O6 keto oxygen atoms of two adjacent guanines directs both its deuterium atoms away from the keto groups. Exocyclic amino groups of guanine (N2) and cytosine (N4) unexpectedly stabilize waters H-bonded to O2 keto oxygens from adjacent cytosines and O6 keto oxygens from adjacent guanines, respectively. Our structure offers the most detailed view to date of DNA solvation in the solid-state undistorted by metal ions or polyamines.

Machine learning for neutron scattering at ORNL*

Machine learning (ML) offers exciting new opportunities to extract more information from scattering data. At neutron scattering user facilities, ML has the potential to help accelerate scientific productivity by empowering facility users with insight into their data which has traditionally been supplied by scattering experts. Such support can help in both speeding up common modeling problems for users, as well as help solve harder problems that are normally time consuming and difficult to address with standard methods. This article explores the recent ML work undertaken at Oak Ridge National Laboratory involving neutron scattering data. We cover materials structure modeling for diffuse scattering, powder diffraction, and small-angle scattering. We also discuss how ML can help to model the response of the instrument more precisely, as well as enable quick extraction of information from neutron data. The application of super-resolution techniques to small-angle scattering and peak extraction for diffraction will be discussed.

Direct detection of coupled proton and electron transfers in human manganese superoxide dismutase

Human manganese superoxide dismutase is a critical oxidoreductase found in the mitochondrial matrix. Concerted proton and electron transfers are used by the enzyme to rid the mitochondria of O₂^•−. The mechanisms of concerted transfer enzymes are typically unknown due to the difficulties in detecting the protonation states of specific residues and solvent molecules at particular redox states. Here, neutron diffraction of two redox-controlled manganese superoxide dismutase crystals reveal the all-atom structures of Mn³⁺ and Mn²⁺ enzyme forms. The structures deliver direct data on protonation changes between oxidation states of the metal. Observations include glutamine deprotonation, the involvement of tyrosine and histidine with altered pK_as, and four unusual strong-short hydrogen bonds, including a low barrier hydrogen bond. We report a concerted proton and electron transfer mechanism for human manganese superoxide dismutase from the direct visualization of active site protons in Mn³⁺ and Mn²⁺ redox states.

Human manganese superoxide dismutase (MnSOD) is an oxidoreductase that uses concerted proton and electron transfers to reduce the levels of superoxide radicals in mitochondria, but mechanistic insights into this process are limited. Here, the authors report neutron crystal structures of Mn³⁺SOD and Mn²⁺SOD, revealing changes in the protonation states of key residues in the enzyme active site during the redox cycle.

Interactive Visual Study of Multiple Attributes Learning Model of X-Ray Scattering Images

Existing interactive visualization tools for deep learning are mostly applied to the training, debugging, and refinement of neural network models working on natural images. However, visual analytics tools are lacking for the specific application of x-ray image classification with multiple structural attributes. In this paper, we present an interactive system for domain scientists to visually study the multiple attributes learning models applied to x-ray scattering images. It allows domain scientists to interactively explore this important type of scientific images in embedded spaces that are defined on the model prediction output, the actual labels, and the discovered feature space of neural networks. Users are allowed to flexibly select instance images, their clusters, and compare them regarding the specified visual representation of attributes. The exploration is guided by the manifestation of model performance related to mutual relationships among attributes, which often affect the learning accuracy and effectiveness. The system thus supports domain scientists to improve the training dataset and model, find questionable attributes labels, and identify outlier images or spurious data clusters. Case studies and scientists feedback demonstrate its functionalities and usefulness.

Opportunities and challenges in neutron crystallography

Neutron and X-ray crystallography are complementary to each other. While X-ray scattering is directly proportional to the number of electrons of an atom, neutrons interact with the atomic nuclei themselves. Neutron crystallography therefore provides an excellent alternative in determining the positions of hydrogens in a biological molecule. In particular, since highly polarized hydrogen atoms (H+) do not have electrons, they cannot be observed by X-rays. Neutron crystallography has its own limitations, mainly due to inherent low flux of neutrons sources, and as a consequence, the need for much larger crystals and for different data collection and analysis strategies. These technical challenges can however be overcome to yield crucial structural insights about protonation states in enzyme catalysis, ligand recognition, as well as the presence of unusual hydrogen bonds in proteins.