Saturday, December 7, 2024
No menu items!
HomeNatureA publishing platform that places code front and centre

A publishing platform that places code front and centre

An open book with a laser beam shinging across the pages

Credit: Getty

Last month, the Microscopy Society of America (MSA) quietly launched a journal that its creators hope will lead academic publishing in the future. Elemental Microscopy, a publication focused on reviews and methods tutorials, leverages an authoring and publishing platform called Curvenote to create rich, interactive, digital papers with coding and data analysis at their heart. By allowing readers to explore data and recreate results in the publication, the firm behind the platform, also called Curvenote, and its growing list of clients seek to ease science’s long-standing reproducibility crisis and modernize the scientific literature.

“Science has changed a lot — both through the explosion of big data and the digitization of science — but fundamentally, we still write papers like we did 100 years ago, in a way that doesn’t prioritize methods and data analysis,” says Colin Ophus, an incoming computational microscopist at Stanford University in California and Elemental Microscopy’s first editor-in-chief. “Through this new partnership, we want to show the whole pipeline, from the experiment to the raw data to the final figure, and all the steps that go into that.” The partnership was announced at the MSA’s annual meeting in Cleveland, Ohio, in July.

Based in Calgary, Canada, Curvenote is a web-first platform built atop a slate of open-source tools that are already widely used by data scientists. Researchers often host their data in GitHub repositories and run analyses in computational notebook environments such as R Markdown or Jupyter Notebook. But to publish a paper, they must distil hundreds of lines of code into a paragraph or two, sacrificing important details and relegating the code itself to supplemental materials. When the paper is converted to a Word document or PDF, it separates text and figures from the code and data used to create them and flattens otherwise dynamic data to static representations. By combining computational notebooks with a user-friendly formatting language called MyST Markdown, Curvenote makes it possible to draft and manage an interactive manuscript from start to finish, including peer review, without the cumbersome translations. And because Curvenote outputs to an XML format called Journal Article Tag Suite, which many publishers use, the tool should allow any publisher to work with the documents directly, says Rowan Cockett, a geophysicist who co-founded the platform in 2019.

Dual benefits

“I think Curvenote is very advanced, and I’m very happy with what they’ve done,” says Alberto Pepe, vice-president of technology at the biotechnology research non-profit organization Sage Bionetworks in Seattle, Washington. Pepe developed a collaborative publishing platform called Authorea before selling it to scientific publisher Wiley in 2018. “It really is a step above.”

Screen recording showing the interactive features of a conference paper hosted on Curvenote

The publishing platform Curvenote is used to create interactive articles, allowing readers to view data and figures easily.Credit: Curvenote (CC-BY-4.0); Moore et al./SciPy Proceedings 2023; Section 3.2.3 (CC-BY-3.0)

According to Cockett, leveraging open-access tools to develop Curvenote has dual benefits: not only do these choices support the company’s mission of advancing open science, but they also give the platform staying power. “It’s very important that the tools we use are being stewarded by the open-science and open-source communities, such that they have deep integration,” he says, noting that past attempts to create computationally reproducible articles using custom tools have often struggled to keep up. “Anytime there’s an update, things break and initiatives fail.”

Curvenote stems from Cockett’s doctoral studies at the University of British Columbia in Vancouver, Canada, when he saw how challenging it was for members of his laboratory to explain their coding to one another, or for others to replicate it after people had moved on. Although the platform began as a way for individual scientists to share their work, Cockett says that the team has also worked with secondary-school students, lab groups, institutions, and, over the past two years, publishers. In addition to the MSA — which plans to announce a call for papers for Elemental Microscopy later this year — Curvenote has partnered with the American Geophysical Union (AGU) and the publications SciPy Proceedings and Physiome.

Alan Lujan, a computational economist at the open-source modelling project Econ-ARK, based at Johns Hopkins University in Baltimore, Maryland, who used Curvenote to create an article under review for SciPy Proceedings, praises the platform for its utility, including its seamless integration with scientific visualization libraries, such as Plotly, Leaflet, Altair and Bokeh.

By launching a virtual environment in the paper, readers can view and execute the underlying code, and edit that code to experiment with the data. Data are hosted in the Google Cloud Platform and executed using Binder, making computation free to access, but also limiting the computing resources available. The interface supports interactive citations, including the ability to cross-reference and automatically number figures, equations and tables. When users hover over these interactive references, pop-ups provide information without requiring the reader to navigate away from the text.

Researchers can author MyST Markdown documents in any plain-text editor, including Jupyter or the Quarto document system. Curvenote also hosts its own authoring and content-management platform that allows collaborators to edit MyST Markdown documents and append comments in a Google Doc-like format, even if they have limited coding experience. “I’ve been working a lot on publishing code and data together, and I find that people can be very hesitant to work with these computational notebooks because it seems very complicated to collaborate,” says Sarah Pederzani, an archaeologist and palaeoclimate scientist at the University of Utah in Salt Lake City. “Anything that can help with the uptake of these products is great.”

At its annual meeting in December 2023, the AGU announced a pilot project called Notebooks Now! in partnership with Curvenote, and released a handful of publications to showcase what it can do. In an accompanying editorial1, the AGU’s leadership said that the initiative would “revolutionize the way data- and computation-rich scientific research is performed and published”. Instead of PDFs, Notebooks Now! would make dynamic computational documents the “version of record”, which could then be exported to PDF and other formats.

According to an AGU spokesperson, the interest from its membership in Notebooks Now! has been “very positive”, and the AGU is now seeking funding to move the project forward “in the next few months”. At first, scientists will be able to submit Curvenote-formatted manuscripts to the journal Earth and Space Science — probably by the end of the year. The goal is to eventually offer the option for all AGU journals.

Kayla Iacovino, an experimental petrologist at the NASA Johnson Space Center in Houston, Texas, who reformatted one of her publications for Notebooks Now!, says that the process was time consuming, but ultimately created a better manuscript. One scientist reached out afterwards, she says, to share that Curvenote had made it possible to ‘experience’ Iacovino’s publication, rather than simply reading it. “It really took it to the next level,” she says.

RELATED ARTICLES

Most Popular

Recent Comments