Artificial Intelligence and Machine Learning in Crystallography Editorial for Crystallography Reviews, Issue 2 of Volume27, 2021

“Everything we love about civilization is a product of intelligence, so amplifying our human intelligence with artificial intelligence has the potential of helping civilization flourish like never before – as long as we manage to keep the technology beneficial.” said Max Tegmark, President of the Future of Life Institute [1]. There is half a century of evolution behind artificial intelligence (AI) and machine learning (ML). The exponentially developing technology can do a good job at narrow tasks for example in mathematics, modelling climate change, internet searches, facial recognition, speech recognition, driving autonomous cars, customer service, playing chess, or Facebook uses algorithms to block content that breaks its rules. It can be applied in automated stock trading, it is offered for the commercial sectors, solving business problems for public and private sectors.

Science fiction often portrays artificial intelligence with human-like characteristics, which emerges conversations about the impact on society and around the ethics of AI. Artificial General or Super Intelligence is a theoretical form of AI, where it would have a self-aware consciousness that had the ability to solve problems surpassing the intelligence and capacity of the human brain. An example is HAL, the rogue computer assistant in 2001: A Space Odyssey. Back to reality, algorithms cannot understand the essence of humans: emotion, morality, culture, since these abilities cannot be expressed in mathematical equations.

Artificial Intelligence enables problem-solving by the combination of computer science and robust datasets. Machine learning is the subfields of artificial intelligence. Deep learning refers to a neural network, which comprise of multiple hidden layers between the input and output layers. Machine learning and deep learning differs in the way how their algorithms learn. Machine learning is more dependent on human intervention, what determines the hierarchy of features. A deep learning-based system would be able to achieve the same task in a much shorter time.

Melanie Vollmar and Gwyndaf Evans from the Diamond Light Source Ltd., and from the Rosalind Franklin Institute, respectively, both in Harwell Science and Innovation Campus, Didcot, UK, review the “Machine learning applications in macromolecular X-ray crystallography” in Issue 2 of Volume 27 of Crystallography Reviews. We can quickly understand why introducing Artificial Intelligence is so much investigated and needed at the Diamond Light Source reading the facts: throughout 2020 user access to MX beamlines was almost exclusively remote, in the 11 months from June 2020 over 33,000 data sets were measured, typically less than 3 minutes being used for each crystal sample, the yearly quantities of measured data reach many Petabyte. We may learn from the article that AI can contribute to the question of crystallisability, to the detection of the presence of crystals in crystallisation trials, to forecast of experimental data and data analysis outcome, to have real-time feedback on the quality of data being measured, to the minimisation of human intervention of data assessment, and also to the protein structure prediction from amino acid sequence data. The review starts with a historical retrospection to the evolution of AI and ML, and presents an overview of concepts and terminology, offers introduction and guide to their application in crystallography and reports examples of their influence on macromolecular crystallography.

The second topical review of this issue is on “Cocrystallization as a tool to stabilize liquid active ingredients” by Alessia Bacchi and Paolo Pio Mazzeo from the Department of Chemistry, Life Sciences and Environmental Sustainability, and Biopharmanet-TEC, respectively, both University of Parma, Italy. The IUPAC has not yet expressed a validated definition of cocrystals. Neither in the ongoing definition of the FDA issued in 2018 nor in the present IUCr statement, there is no mention of the physical state of the cocrystal formers at ambient conditions, however IUCr refers that cocrystals are neither solvates nor simple salts. In the cocrystal formation with liquid active ingredients, such as pharmaceuticals, agrochemicals and nutraceuticals, it can be crucial for their application to influence their physico-chemical properties, like thermal stability, solubility, hygroscopicity in order to improve application, manufacturing, transportation, and storage properties. The design of cocrystals requires the deep understanding of supramolecular interactions, and the application of its principles. The article provides an introduction to the basic energetic considerations on the propensity of cocrystallization of liquid ingredients, analyses the structural requirements for their cocrystallization, and presents a range of examples, where cocrystallization has provided a tool for stabilizing liquid active ingredients in a solid form.

This issue of Crystallography Reviews offers three book reviews. David Keen from ISIS Facility, Rutherford Appleton Laboratory, Harwell Campus, Didcot, Oxfordshire, UK, has read the book of Frank Hoffmann with the title of “Introduction to Crystallography” published by Springer International Publishing, Switzerland in 2020. There is no better way to advertise a publication than the confession “I have read (it) from cover to cover”, while the reviewer shares personal experiences in crystallography related to the book. The original German title “Faszination Kristalle and Symmetrie” (Springer Spektrum, 2016) is different, anyhow both titles express the aim of the author to give the basic concepts clearly and gently, and without detailed mathematics to those researcher who use crystallography as a tool.

Claire Murray from Diamond Light Source Ltd., Oxfordshire, UK has prepared an unusual book review on an unusual book: “Department of mind-blowing theories” by Tom Gauld, Edinburgh, Drawn and Quarterly, Canongate Books, Publishers Group UK, 2020. As the reviewer writes, the comic book explores the (sur)real life of a scientist in all its gore and glory, referring a wide variety of griefs that may happen in a laboratory, presenting the not necessarily gently everyday life of a researcher fighting for or with collaborations, presentations, paper preparations, having pitfalls in the contact with media, etc., throughout filled with humour and completed with artwork of simple and clean design and colour.

The co-authors of the third book, reviewed by Ernst Ferg from Nelson Mandela University, Gqeberha (formerly Port Elizabeth), South Africa, come from the field of science and engineering and arrive from both academia and the industry. “Powering the U.S. army of the future” was written by contributors of the Committee on Powering the U.S. Army of the Future; Board on Army Research and Development; Division on Engineering and Physical Sciences; National Academies of Sciences, Engineering, and Medicine, Washington, DC, published by The National Academies Press, United States in 2021. The book is recommended those, who are working in material science, on materials used in energy storage devices. There are short, easy to read chapters with clear definitions of terminologies that might be specific to the related technology. We learn that within the US military, the priorities are for the support and simplification of energy logistics. Fuel and engine research are presented. Modern energy technologies are also described as of secondary interest, like use of remote electrolysers, the possibility of using Solid Oxide Fuel Cells, and the idea of using small portable micro-grid nuclear power systems. The book provides significant information around the use of Li-ion battery and related next generation batteries, such as other energy storage Zn-based chemistries, as well a thermophotovoltaic system using nanophotonic material is also discussed all as emerging energy related technologies.

As ever we welcome new ideas for review articles, and for suggestions regarding books to be reviewed. Please contact me at the e-mail address below.

I look forward to welcoming your submissions.

 

Original post: https://www.tandfonline.com/doi/full/10.1080/0889311X.2021.2000094

Leave a Reply

Your email address will not be published. Required fields are marked *