Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Access to clean, affordable and reliable energy has been a cornerstone of the world's increasing prosperity and economic growth since the beginning of the industrial revolution. Our use of energy in the twenty–first century must also be sustainable. Solar and water–based energy generation, and engineering of microbes to produce biofuels are a few examples of the alternatives. This Perspective puts these opportunities into a larger context by relating them to a number of aspects in the transportation and electricity generation sectors. It also provides a snapshot of the current energy landscape and discusses several research and development opportunities and pathways that could lead to a prosperous, sustainable and secure energy future for the world.

Opportunities and challenges for a sustainable energy future

Although fire is now rarely used in synthetic chemistry, it was not until Robert Bunsen invented the burner in 1855 that the energy from this heat source could be applied to a reaction vessel in a focused manner. The Bunsen burner was later superseded by the isomantle, oil bath, or hot plate as a source for applying heat to a chemical reaction. In the past few years, heating and driving chemical reactions by microwave energy has been an increasingly popular theme in the scientific community. This nonclassical heating technique is slowly moving from a laboratory curiosity to an established technique that is heavily used in both academia and industry. The efficiency of "microwave flash heating" in dramatically reducing reaction times (from days and hours to minutes and seconds) is just one of the many advantages. This Review highlights recent applications of controlled microwave heating in modern organic synthesis, and discusses some of the underlying phenomena and issues involved.

https://onlinelibrary.wiley.com/doi/pdf/10.1002/anie.200400655

Controlled microwave heating in modern organic synthesis.

Over the past few decades, researchers have established artificial enzymes as highly stable and low-cost alternatives to natural enzymes in a wide range of applications. A variety of materials including cyclodextrins, metal complexes, porphyrins, polymers, dendrimers and biomolecules have been extensively explored to mimic the structures and functions of naturally occurring enzymes. Recently, some nanomaterials have been found to exhibit unexpected enzyme-like activities, and great advances have been made in this area due to the tremendous progress in nano-research and the unique characteristics of nanomaterials. To highlight the progress in the field of nanomaterial-based artificial enzymes (nanozymes), this review discusses various nanomaterials that have been explored to mimic different kinds of enzymes. We cover their kinetics, mechanisms and applications in numerous fields, from biosensing and immunoassays, to stem cell growth and pollutant removal. We also summarize several approaches to tune the activities of nanozymes. Finally, we make comparisons between nanozymes and other catalytic materials (other artificial enzymes, natural enzymes, organic catalysts and nanomaterial-based catalysts) and address the current challenges and future directions (302 references).

Nanomaterials with enzyme-like characteristics (nanozymes): next-generation artificial enzymes

In the investigation of efficient chemical transformations, the carbon-carbon bond-forming reactions play an outstanding role. In this context, organocatalytic processes have achieved considerable attention. 1 Beside their facile reaction course, selectivity, and environmental friendliness, new synthetic strategies are made possible. Particularly, the inversion of the classical reactivity (Umpolung) opens up new synthetic pathways. 2 In nature, the coenzyme thiamine (vitamin B1), a natural thiazolium salt, utilizes a catalytic variant of this concept in biochemical processes as nucleophilic acylations. 3 The catalytically active species is a nucleophilic carbene. 4

Organocatalysis by N-Heterocyclic Carbenes

Enzymes are increasingly being used as biocatalysts in the generation of products that have until now been derived using traditional chemical processes. Such products range from pharmaceutical and agrochemical building blocks to fine and bulk chemicals and, more recently, components of biofuels. For a biocatalyst to be effective in an industrial process, it must be subjected to improvement and optimization, and in this respect the directed evolution of enzymes has emerged as a powerful enabling technology. Directed evolution involves repeated rounds of (i) random gene library generation, (ii) expression of genes in a suitable host and (iii) screening of libraries of variant enzymes for the property of interest. Both in vitro screening–based methods and in vivo selection–based methods have been applied to the evolution of enzyme function and properties. Significant developments have occurred recently, particularly with respect to library design, screening methodology, applications in synthetic transformations and strategies for the generation of new enzyme function.

Directed evolution drives the next generation of biocatalysts

Aliphatic hydrocarbons such as fatty alcohols and petroleum-derived alkanes have numerous applications in the chemical industry. In recent years, the renewable synthesis of aliphatic hydrocarbons has been made possible by engineering microbes to overaccumulate fatty acids. However, to generate end products with the desired physicochemical properties (e.g., fatty aldehydes, alkanes, and alcohols), further conversion of the fatty acid is necessary. A carboxylic acid reductase (CAR) from Mycobacterium marinum was found to convert a wide range of aliphatic fatty acids (C6–C18) into corresponding aldehydes. Together with the broad-substrate specificity of an aldehyde reductase or an aldehyde decarbonylase, the catalytic conversion of fatty acids to fatty alcohols (C8–C16) or fatty alkanes (C7–C15) was reconstituted in vitro. This concept was applied in vivo, in combination with a chain-length-specific thioesterase, to engineer Escherichia coli BL21(DE3) strains that were capable of synthesizing fatty alcohols and alkanes. A fatty alcohol titer exceeding 350 mg·L−1 was obtained in minimal media supplemented with glucose. Moreover, by combining the CAR-dependent pathway with an exogenous fatty acid-generating lipase, natural oils (coconut oil, palm oil, and algal oil bodies) were enzymatically converted into fatty alcohols across a broad chain-length range (C8–C18). Together with complementing enzymes, the broad substrate specificity and kinetic characteristics of CAR opens the road for direct and tailored enzyme-catalyzed conversion of lipids into user-ready chemical commodities.

https://www.pnas.org/content/pnas/110/1/87.full.pdf

Carboxylic acid reductase is a versatile enzyme for the conversion of fatty acids into fuels and chemical commodities

Enantiomerically pure chiral amines are valuable building blocks for the synthesis of pharmaceutical drugs and agrochemicals. Indeed it is estimated that currently 40 % of pharmaceuticals contain a chiral amine component in their structure. Chiral amines are also widely used as resolving agents for diastereomeric salt crystallization. One of the challenges of preparing chiral amines in enantiomerically pure form is the development of cost-effective and sustainable catalytic methods that are able to address the requirement for the entire range of primary, secondary and tertiary amines. In this review we highlight various biocatalytic strategies that have been developed, particularly those based upon asymmetric synthesis or their equivalent therefore (i.e. dynamic kinetic resolution, deracemisation) in which yields and enantiomeric excesses approaching 100 % can be attained. Particular attention is given to the use of monoamine oxidase (MAO-N) from Aspergillus niger which has been engineered by directed evolution to provide a tool-box of variants which can generate enantiomerically pure primary, secondary and tertiary amines. These MAO-N variants are combined with non-selective chemical reducing agents in deracemisation processes.

Biocatalytic Approaches to the Synthesis of Enantiomerically Pure Chiral Amines

α-Chiral amines are key intermediates for the synthesis of a plethora of chemical compounds at industrial scale. We present a biocatalytic hydrogen-borrowing amination of primary and secondary alcohols that allows for the efficient and environmentally benign production of enantiopure amines. The method relies on a combination of two enzymes: an alcohol dehydrogenase (from Aromatoleum sp., Lactobacillus sp., or Bacillus sp.) operating in tandem with an amine dehydrogenase (engineered from Bacillus sp.) to aminate a structurally diverse range of aromatic and aliphatic alcohols, yielding up to 96% conversion and 99% enantiomeric excess. Primary alcohols were aminated with high conversion (up to 99%). This redox self-sufficient cascade possesses high atom efficiency, sourcing nitrogen from ammonium and generating water as the sole by-product.

/pdf/conversion-of-alcohols-to-enantiopure-amines-through-dual-icdpnzj1aj.pdf

Conversion of alcohols to enantiopure amines through dual-enzyme hydrogen-borrowing cascades.

The combination of sequential biocatalytic reactions, via non-natural synthetic cascades, is a rapidly developing field and leads to the generation of complex valuable chemicals from simple precursors. As the toolbox of available biocatalysts continues to expand, so do the options for biocatalytic retrosynthesis of a target molecule, leading to alternative routes employing enzymatic transformations. The implementation of such cascade reactions requires careful consideration, particularly with respect to whether the pathway is constructed in vitro or in vivo. In this Perspective, we describe the relative merits of in vitro, in vivo, and hybrid approaches to building biocatalytic cascades and showcase recent developments in the area. We also highlight the factors that influence the design and implementation of purely enzymatic or chemoenzymatic, one-pot, multistep pathways.

Nicholas J. Turner

Papers

Directed evolution drives the next generation of biocatalysts

Carboxylic acid reductase is a versatile enzyme for the conversion of fatty acids into fuels and chemical commodities

Biocatalytic Approaches to the Synthesis of Enantiomerically Pure Chiral Amines

Conversion of alcohols to enantiopure amines through dual-enzyme hydrogen-borrowing cascades.

Constructing Biocatalytic Cascades: In Vitro and in Vivo Approaches to de Novo Multi-Enzyme Pathways