Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Black box machine learning models can be dangerous for high-stakes decisions. They rely on untrustworthy databases, and their predictions are difficult to troubleshoot, explain and error check for real-time predictions. Their use leads to serious ethics and accountability issues.
Lokwani et al. discuss the necessary considerations when performing spectral cytometry on highly autofluorescent samples to extract phenotypic information from autofluorescence spectra and perform accurate quantification of fluorescent labels.
Addressing current and future global problems requires sustainable practices in chemistry. We discuss ways to incorporate sustainability in laboratory practice, processes and beyond.
Laboratories have a large environmental impact, with high levels of resource consumption and waste generation. In this article, the author discusses some of the actionable strategies that can bring real and impactful improvements, encompassing education, community engagement and the adoption of best practices by researchers. Building a global culture of sustainability in science will be crucial to reducing the carbon footprint of laboratories.
The scaling up of a chemical reaction is a complex process. Chemists should pay special attention to a number of key factors, including the choice of route, reagents and solvents; health and safety considerations; the isolation and purification of the desired product; and the development of robust supporting analytical methodology.
X-ray induced structural damage is well known, but the potential for changes in the kinetics of physical and chemical processes is rarely recognized or considered. These can happen over a wide intensity range, are difficult to predict and often escape detection. The problem deserves more attention from experimentalists.
COVID-19 has resulted in long-term effects on science and research. The way in which we carry out research has had to rapidly adapt as a result of the pressures placed on scientists, leading to the development of innovative approaches to research.
A large sample size, or N, increases the sensitivity of an experiment to detect differences between treatment groups. However, the biological entity that N refers to may not be obvious. Defining the wrong entity can inflate the sample size and increase both false-positive and false-negative results.
Orphan drug development is a rapidly expanding field. Nevertheless, clinical trials for rare diseases can present inherent challenges. Optimal study design and partnerships between academia and industry are therefore required for the successful development, delivery and clinical approval of effective therapies in this group of disorders.
Automated single-particle picking in electron cryo-microscopy data has seen important advances in the past couple of years and now enables computer-assisted particle selection even for challenging datasets. These advances have implications for streamlined and automated image processing, with potential benefits for improving the resolution of resulting structures.
Ensuring reproducibility and replicability has been an issue in many scientific disciplines in the past decade. Here, we discuss another ‘R’ that has not gotten enough airtime — reanalysis. We cover how open science and a focus on enabling reanalysis also make the goals of reproducibility and replicability easier to achieve.
