The extensive use of graphs in chemistry to model both reactions and molecules creates challenges for machine learning. Most of the previous and ongoing research in the field solves the problem by translating molecular graphs into forms that fit cleanly within existing paradigms and tools: images; text; and binary fingerprints. Graph representations themselves seem out-of-place at first.

InChI is a software package that generates unique chemical identifiers. Although InChI’s C source yields executables for many platforms, there are exceptions. For a long time, the Web browser was one of them. A few years ago, I showed how to compile InChI to JavaScript. That solution works, but recent developments, both in browsers and InChI itself, suggest a better approach is possible.

Cheminformatics concerns itself with two entities: molecules and reactions. A molecule embodies chemistry’s static nature, whereas a reaction embodies the dynamic. Although there are important problems to solve on both sides, cheminformatics tends to focus on molecular tools over reaction tools. One reason is complexity. Reaction tools require solutions to all of the problem required by molecule tools, and then some. Then there’s graph theory.

Graphs pervade science and technology. As such, many kinds of software projects rely on a graph library. This article introduces Gamma, a new graph library written in Rust. Motivation Graphs are used heavily within cheminformatics , a field concerned with the collection, storage, and retrieval of information about substances. Most cheminformatics systems use a library for manipulating molecular structures.

NOTE: a follow-up article is available. Aromaticity pervades organic chemistry. Not surprisingly, it pops up in cheminformatics as well. But like many ideas in chemistry, reduction to software has a way of exposing hidden complexity. The SMILES language offers a case in point. Here the fuzzy notion of aromaticity collides with the rigor required by an information exchange format.

Graphs are ubiquitous in science and technology. As such, many software projects use graphs in one form or another. Here I discuss Petgraph, a general-purpose graph library written in Rust. The main features of Petgraph are illustrated with short code samples. About Petgraph Started in 2014, Petgraph is Rust’s most popular graph library. According to crates.io, Petgraph has been downloaded over 2.1 million times.

Rust is a safe systems programming language. Although C and C++ are systems languages, they’re not safe. Specifically, Rust is a “type safe language”, meaning that the compiler ensures that every program has well-defined behavior. Although other languages make the same guarantee, Rust does so without a garbage collector, runtime, or manual memory management.

Few decisions are more important in a software project than programming language. The choice often boils down to four factors: maturity; libraries; platforms; and tooling. For well-established languages, the choice is sometimes easy: JavaScript for client-side web development; Python for machine learning and numerical analysis; C/C++ for embedded systems and operating systems; and Java for enterprise applications.

WebAssembly (aka Wasm) is a new W3C recommendation and the second target language to be supported by all browsers. WebAssembly is also fast becoming a full-blown runtime environment outside the browser. Unlike JavaScript, WebAssembly was designed with fast execution in mind. A previous article described how to compile C directly to WebAssembly.

Graphs are ubiquitous data structures in computing, appearing in domains ranging from networks to logistics to chemistry. Despite these diverse applications, relatively little has been said about the irreducible elements of graph-like behavior. This article introduces a minimal application programming interface (API) for graphs that I’ve developed over the last several years.

What’s the difference between computational chemistry and cheminformatics? Computational chemists encode all the hydrogens. This tongue-in-cheek statement, whose origins I don’t recall, isn’t far off the mark. It stands to reason that if hydrogens are going to be deleted, there needs to be a good way to bring them back. This article proposes some language intended to simplify what has become an unnecessarily complex problem.