Distributed systems pervade modern computer technology. Examples include multi-core CPUs, GPUs, computer clusters, and the Internet itself. Although many forms of distributed computing have been developed, their main purpose is to increase the utility of individual resources through cooperation, often involving some form of parallelization.

Recent advances in machine learning techniques have yielded systems that meet or even exceed human pattern-recognition capability. These powerful techniques are now starting to be used for chemical structure-property prediction. This article highlights a new approach that, in a break with past systems, works in a way that will be immediately (perhaps uncannily) recognizable to many research chemists.

Until recently, large databases of machine-readable chemical reactions were rare, constrained in their allowed uses, and extremely expensive. That changed in 2014 with the publication of a dataset of organic chemical reactions extracted from US patents and patent applications. What follows is an introduction to this dataset and a discussion of some of the ways it has been used.

Writing a parser from scratch can be a daunting task. For a beginner, the topic may seem barricaded behind a thick wall of theory and vocabulary. Getting to the heart of the matter, running code, can take a lot of time and effort. This article describes a shift in thinking that can significantly reduce the time needed to get started and finish.

The ES Module standard defines a way to build and assemble JavaScript modules. Currently supported in modern browsers and Node.js for many months now, ES Modules will over time render various ad-hoc module systems currently in use obsolete. Given their recent emergence from alpha, however, support for ES Modules can vary across tools.

A fingerprint is a useful kind of low-resolution molecular representation. Fingerprints come in a many forms and have been applied to a diverse range of problems over the last few decades. Extended Connectivity Fingerprints (ECFPs, or “Circular Fingerprints”) offer a number of advantages over other schemes. This article describes the structure and generation of ECFPs at a high level.

Technological change has a way of making fools out of experts and experts out of fools. One of the reasons that evaluating the future impact of new technologies is so difficult is that those with the greatest potential often start out looking - and acting - like junk. They can’t initially do the job nearly as well as the technologies they end up replacing. They are written off as toys.

Riders of commuter trains are accustomed to using maps like those produced by the London Underground: These maps are capable of condensing large amounts of information into a useful package. Transit map representations, known more generally as graph or network representations, are especially good at revealing the big picture, although their applications are numerous and can be found these days just about anywhere you find large datasets.

Although not widely-known, the American Chemical Society permits unlimited, free article downloads across its vast journal lineup without a subscription thanks to the Articles on Request program (AoR). Of course there is a catch, explained below, but the official ACS Publications policy is that a significant portion of its collection can in principle be downloaded by anyone for free.

A previous post detailed my failed attempt to test a rather interesting if not well-known program offered by ACS to corresponding publication authors: “free access” to final published articles after 12 months under the Articles on Request (AoR) program. In that post I made a mistake that I’ll attempt to correct here.

Please see the revised post, A Third Failed Test of ACS Articles on Request and How to Help . Although the introduction, conclusions and comments section of the post below may be useful, the links and discussion around them should be disregarded. A recent post described the lack of public author-oriented documentation for the ACS Articles on Request (AoR) program.