Establishing a training set through the visual analysis of crystallization trials. Part I: approximately 150,000 images

20 years of crystal hits: progress and promise in ultrahigh-throughput crystallization screening

Diffraction-based structural methods contribute a large fraction of the biomolecular structural models available, providing a critical understanding of macromolecular architecture. These methods require crystallization of the target molecule, which remains a primary bottleneck in crystal-based structure determination. The National High-Throughput Crystallization Center at Hauptman-Woodward Medical Research Institute has focused on overcoming obstacles to crystallization through a combination of robotics-enabled high-throughput screening and advanced imaging to increase the success of finding crystallization conditions. This paper will describe the lessons learned from over 20 years of operation of our high-throughput crystallization services. The current experimental pipelines, instrumentation, imaging capabilities and software for image viewing and crystal scoring are detailed. New developments in the field and opportunities for further improvements in biomolecular crystallization are reflected on.

Polo: an open-source graphical user interface for crystallization screening

A multi-platform open-source Python-based graphical user interface has been developed to provide access to automated classification and data management tools for biomolecular crystallization screening.

Polo is a Python-based graphical user interface designed to streamline viewing and analysis of images to monitor crystal growth, with a specific target to enable users of the High-Throughput Crystallization Screening Center at Hauptman-Woodward Medical Research Institute (HWI) to efficiently inspect their crystallization experiments. Polo aims to increase efficiency, reducing time spent manually reviewing crystallization images, and to improve the potential of identifying positive crystallization conditions. Polo provides a streamlined one-click graphical interface for the Machine Recognition of Crystallization Outcomes (MARCO) convolutional neural network for automated image classification, as well as powerful tools to view and score crystallization images, to compare crystallization conditions, and to facilitate collaborative review of crystallization screening results. Crystallization images need not have been captured at HWI to utilize Polo’s basic functionality. Polo is free to use and modify for both academic and commercial use under the terms of the copyleft GNU General Public License v3.0.

Cinder: keeping crystallographers app-y

The process of producing suitable crystals for X-ray diffraction analysis most often involves the setting up of hundreds (or thousands) of individual crystallization trials, each of which must be repeatedly examined for crystals or hints of crystallinity. Currently, the only real way to address this bottleneck is to use an automated imager to capture images of the trials. However, the images still need to be assessed for crystals or other outcomes. Ideally, there would exist some rapid and reliable machine-analysis tool to translate the images into a quantitative result. However, as yet no such tool exists in wide usage, despite this being a well recognized problem. One of the issues in creating robust automatic image-analysis software is the lack of reliable data for training machine-learning algorithms. Here, a mobile application, Cinder, has been developed which allows crystallization images to be scored quickly on a smartphone or tablet. The Cinder scores are inserted into the appropriate table in a crystallization database and are immediately available to the user through a more sophisticated web interface, allowing more detailed analyses. A sharp increase in the number of scored images was observed after Cinder was released, which in turn provides more data for training machine-learning tools.

Classification of crystallization outcomes using deep convolutional neural networks

The Machine Recognition of Crystallization Outcomes (MARCO) initiative has assembled roughly half a million annotated images of macromolecular crystallization experiments from various sources and setups. Here, state-of-the-art machine learning algorithms are trained and tested on different parts of this data set. We find that more than 94% of the test images can be correctly labeled, irrespective of their experimental origin. Because crystal recognition is key to high-density screening and the systematic analysis of crystallization experiments, this approach opens the door to both industrial and fundamental research applications.

Automated Protocols for Macromolecular Crystallization at the MRC Laboratory of Molecular Biology

When high quality crystals are obtained that diffract X-rays, the crystal structure may be solved at near atomic resolution. The conditions to crystallize proteins, DNAs, RNAs, and their complexes can however not be predicted. Employing a broad variety of conditions is a way to increase the yield of quality diffraction crystals. Two fully automated systems have been developed at the MRC Laboratory of Molecular Biology (Cambridge, England, MRC-LMB) that facilitate crystallization screening against 1,920 initial conditions by vapor diffusion in nanoliter droplets. Semi-automated protocols have also been developed to optimize conditions by changing the concentrations of reagents, the pH, or by introducing additives that potentially enhance properties of the resulting crystals. All the corresponding protocols will be described in detail and briefly discussed. Taken together, they enable convenient and highly efficient macromolecular crystallization in a multi-user facility, while giving the users control over key parameters of their experiments.

MPI tray: a versatile crystallization plate for membrane proteins

High-throughput crystallization of biological macromolecules is usually performed on multi-well plates, the design of which needs to address different and sometimes conflicting requirements. In this regard, handling of membrane proteins presents a particular challenge owing to the common use of detergents with associated effects on surface tension. Reported here is the design of a new crystallization plate, termed the MPI tray, which is optimized for UV and visible imaging with membrane protein samples. Following basic considerations regarding geometry and material, the surface properties of the plate were subjected to extensive analysis and modification in order to improve the performance in a robotic environment. An electrostatic surface potential was identified as the major problem affecting the automated setup of experiments, and it was found that treatment of the crystallization plate with ethanol is effective in removing this potential.

Computational crystallization

Crystallization is a key step in macromolecular structure determination by crystallography. While a robust theoretical treatment of the process is available, due to the complexity of the system, the experimental process is still largely one of trial and error. In this article, efforts in the field are discussed together with a theoretical underpinning using a solubility phase diagram. Prior knowledge has been used to develop tools that computationally predict the crystallization outcome and define mutational approaches that enhance the likelihood of crystallization. For the most part these tools are based on binary outcomes (crystal or no crystal), and the full information contained in an assembly of crystallization screening experiments is lost. The potential of this additional information is illustrated by examples where new biological knowledge can be obtained and where a target can be sub-categorized to predict which class of reagents provides the crystallization driving force. Computational analysis of crystallization requires complete and correctly formatted data. While massive crystallization screening efforts are under way, the data available from many of these studies are sparse. The potential for this data and the steps needed to realize this potential are discussed.

Identifying, studying and making good use of macromolecular crystals

As technology advances, the crystal volume that can be used to collect useful X-ray diffraction data decreases. The technologies available to detect and study growing crystals beyond the optical resolution limit and methods to successfully place the crystal into the X-ray beam are discussed.

Structural biology has contributed tremendous knowledge to the understanding of life on the molecular scale. The Protein Data Bank, a depository of this structural knowledge, currently contains over 100 000 protein structures, with the majority stemming from X-ray crystallography. As the name might suggest, crystallography requires crystals. As detectors become more sensitive and X-ray sources more intense, the notion of a crystal is gradually changing from one large enough to embellish expensive jewellery to objects that have external dimensions of the order of the wavelength of visible light. Identifying these crystals is a prerequisite to their study. This paper discusses developments in identifying these crystals during crystallization screening and distinguishing them from other potential outcomes. The practical aspects of ensuring that once a crystal is identified it can then be positioned in the X-ray beam for data collection are also addressed.

Statistical analysis of crystallization database links protein physico-chemical features with crystallization mechanisms

X-ray crystallography is the predominant method for obtaining atomic-scale information about biological macromolecules. Despite the success of the technique, obtaining well diffracting crystals still critically limits going from protein to structure. In practice, the crystallization process proceeds through knowledge-informed empiricism. Better physico-chemical understanding remains elusive because of the large number of variables involved, hence little guidance is available to systematically identify solution conditions that promote crystallization. To help determine relationships between macromolecular properties and their crystallization propensity, we have trained statistical models on samples for 182 proteins supplied by the Northeast Structural Genomics consortium. Gaussian processes, which capture trends beyond the reach of linear statistical models, distinguish between two main physico-chemical mechanisms driving crystallization. One is characterized by low levels of side chain entropy and has been extensively reported in the literature. The other identifies specific electrostatic interactions not previously described in the crystallization context. Because evidence for two distinct mechanisms can be gleaned both from crystal contacts and from solution conditions leading to successful crystallization, the model offers future avenues for optimizing crystallization screens based on partial structural information. The availability of crystallization data coupled with structural outcomes analyzed through state-of-the-art statistical models may thus guide macromolecular crystallization toward a more rational basis.

Lessons from high-throughput protein crystallization screening: 10 years of practical experience

INTRODUCTION

X-ray crystallography provides the majority of our structural biological knowledge at a molecular level and, in terms of pharmaceutical design, is a valuable tool to accelerate discovery. It is the premier technique in the field, but its usefulness is significantly limited by the need to grow well-diffracting crystals. It is for this reason that high-throughput crystallization has become a key technology that has matured over the past 10 years through the field of structural genomics. Areas covered : The authors describe their experiences in high-throughput crystallization screening in the context of structural genomics and the general biomedical community. They focus on the lessons learnt from the operation of a high-throughput crystallization-screening laboratory, which to date has screened over 12,500 biological macromolecules. They also describe the approaches taken to maximize the success while minimizing the effort. Through this, the authors hope that the reader will gain an insight into the efficient design of a laboratory and protocols to accomplish high-throughput crystallization on a single-, multiuser laboratory or industrial scale. Expert opinion : High-throughput crystallization screening is readily available but, despite the power of the crystallographic technique, getting crystals is still not a solved problem. High-throughput approaches can help when used skillfully; however, they still require human input in the detailed analysis and interpretation of results to be more successful.

What's in a drop? Correlating observations and outcomes to guide macromolecular crystallization experiments

Observations of crystallization experiments are classified as specific outcomes and integrated through a phase diagram to visualize solubility and thereby direct subsequent experiments. Specific examples are taken from our high-throughput crystallization laboratory which provided a broad scope of data from 20 million crystallization experiments on 12,500 different biological macromolecules. The methods and rationale are broadly and generally applicable in any crystallization laboratory. Through a combination of incomplete factorial sampling of crystallization cocktails, standard outcome classifications, visualization of outcomes as they relate chemically and application of a simple phase diagram approach we demonstrate how to logically design subsequent crystallization experiments.

Crystal cookery - using high-throughput technologies and the grocery store as a teaching tool

Crystallography is a multidisciplinary field that links divergent areas of mathematics, science and engineering to provide knowledge of life on an atomic scale. Crystal growth, a key component of the field, is an ideal vehicle for education. Crystallization has been used with a 'grocery store chemistry' approach and linked to high-throughput remote-access screening technologies. This approach provides an educational opportunity that can effectively teach the scientific method, readily accommodate different levels of educational experience, and reach any student with access to a grocery store, a post office and the internet. This paper describes the formation of the program through the students who helped develop and prototype the procedures. A summary is presented of the analysis and preliminary results and a description given of how the program could be linked with other aspects of crystallography. This approach has the potential to bridge the gap between students in remote locations and with limited funding, and access to scientific resources, providing students with an international-level research experience.

Protein crystallization analysis on the World Community Grid

We have developed an image-analysis and classification system for automatically scoring images from high-throughput protein crystallization trials. Image analysis for this system is performed by the Help Conquer Cancer (HCC) project on the World Community Grid. HCC calculates 12,375 distinct image features on microbatch-under-oil images from the Hauptman-Woodward Medical Research Institute's High-Throughput Screening Laboratory. Using HCC-computed image features and a massive training set of 165,351 hand-scored images, we have trained multiple Random Forest classifiers that accurately recognize multiple crystallization outcomes, including crystals, clear drops, precipitate, and others. The system successfully recognizes 80% of crystal-bearing images, 89% of precipitate images, and 98% of clear drops.

Establishing a training set through the visual analysis of crystallization trials. Part II: crystal examples

As part of a training set for automated image analysis, crystallization screening experiments for 269 different macromolecules were visually analyzed and a set of crystal images extracted. Outcomes and trends are analyzed.

In the automated image analysis of crystallization experiments, representative examples of outcomes can be obtained rapidly. However, while the outcomes appear to be diverse, the number of crystalline outcomes can be small. To complement a training set from the visual observation of 147 456 crystallization outcomes, a set of crystal images was produced from 106 and 163 macromolecules under study for the North East Structural Genomics Consortium (NESG) and Structural Genomics of Pathogenic Protozoa (SGPP) groups, respectively. These crystal images have been combined with the initial training set. A description of the crystal-enriched data set and a preliminary analysis of outcomes from the data are described.