BEYOND GOOGLE DOCS a Small but Growing Suite of Tools Allows Researchers to Author and Edit Scientific Documents As a Team, No E-Mail Required
Total Page：16
File Type：pdf, Size：1020Kb
Work / Technology & tools ILLUSTRATION BY THE PROJECT TWINS THE PROJECT BY ILLUSTRATION COLLABORATIVE WRITING: BEYOND GOOGLE DOCS A small but growing suite of tools allows researchers to author and edit scientific documents as a team, no e-mail required. By Jeffrey M. Perkel raft scientific manuscripts are ethos out for herself. “I wanted to walk the walk Baltimore, Maryland, says it was a “funny typically confidential. So, when Elana of open science,” Botvinnik says. experience” to be tweeted an unpublished Fertig was asked to take a look at an Botvinnik managed her paper as if it were paper. “It’s a very different way of writing than in-development paper on a functional open-source software. She wrote it in a plain- the traditional academic science of not putting gene-annotation strategy, she text editor and placed text files alongside it out before it’s a finished product.” Dexpected to receive the file in a private e-mail. data sets and code for generating figures on Botvinnik’s manuscript was just a shell at this What she got was a public announcement, the code-sharing site GitHub. She invited her point: two of the figures were placeholders, shared on Twitter. four co-authors to submit edits using Git, soft- and the methods section read, “We did things.” The paper had been written by Olga ware that tracks precisely how and when a file But, she says, the fact that the draft was pub- Botvinnik, a computational biologist at the has been changed. And she used a dedicated licly accessible made it easy to solicit feedback Chan Zuckerberg Biohub in San Francisco, tool called Manubot to render the document from co-authors and the broader community. California, who is an advocate of the global as a user-friendly manuscript, which she then “It’s definitely been very, very helpful to be able movement to make research more accessible. In published online and tweeted to the world. to show someone, ‘Here’s what I’m thinking so November 2019, as Botvinnik started preparing Fertig, a computational biologist at far. Here are some figures; here’s some text. her paper, she decided to try this open-science the Johns Hopkins School of Medicine in What do you think?’” 154 | Nature | Vol 580 | 2 April 2020 ©2020 Spri nger Nature Li mited. All ri ghts reserved. ©2020 Spri nger Nature Li mited. All ri ghts reserved. Say ‘collaborative writing’ and most articles, and bundle figures together with written grant applications using Manubot, and researchers probably think of Google Docs, the data used to create them — such features is comfortable with GitHub. But she won’t be the ubiquitous word processor that allows support computational reproducibility. “The using Manubot to write collaborative papers, multiple authors to co-edit a document online intention is that you can create dynamic because the level of programming it involves in real time. But Google Docs lacks features representations of your work, which include tends to be beyond the reach of her clinician that some scientists require, such as refer- code, data and figures, and the narrative, co-authors. “There’s no way they have the ence management, support for code and data all versioned together,” says Matias Piipari, bandwidth to pick up Manubot,” she says. and the ability to directly submit articles to founder of Manuscripts.io, which (like Autho- journals and preprint servers. rea) is now owned by the publisher Wiley. Easing submission Manubot is one of a small but growing Increasingly, developers are fitting these tools number of tools specifically designed for col- “As such tools gain traction, with features to better encapsulate the scien- laborative writing; others include Overleaf, tific process. Some, for instance, support JATS Authorea, Fidus Writer and Manuscripts.io. scientific articles become XML, a file format commonly used in scientific These tools not only close some of the key fea- ever more dynamic.” publishing. ture gaps, but also provide a glimpse of where JATS XML is a structured, semantic file format scientific communication might move next. that provides a rich set of metadata tags for For those who prefer Google Docs, New article elements such as author names, arti- Partners in editing Zealand-based Stencila is developing a plug-in cle sections, funding sources and database Most collaborative writing tools offer that allows authors to enhance documents with accession numbers. Giuliano Maciocci, head of researchers a range of useful functions. Team executable code blocks, data tables and equa- product and user experience at the open-access members can keep documents private or share tions. Based on steganography, a cryptographic journal eLife, explains that the format “decou- them with select collaborators; track changes trick in which data are encoded in images, ples the structure of the article from its pres- and comment on the text; and edit documents Stencila’s plug-in was written to “bridge that entation”, which makes the data easier to simultaneously or asynchronously with their gap between the coders and the clickers”, says search, access and manipulate. collaborators. founder Nokome Bentley. “It’s taking the code Editors typically build documents by Science-focused programs supplement those to the environment that clickers are used to.” converting author-submitted files into a for- with features aimed at the research community, mat they can publish in, Maciocci says — a such as built-in citation management. (Some Coder workflows labour-intensive, error-prone process. To help citation managers can integrate with Google Manubot, by contrast, tends to appeal to automate this process, eLife is developing a tool Docs using plug-ins, such as Zotero and Paper- coders. Developed in the laboratory of bio- called Libero Editor, which it hopes to release pile.) Users can generally import libraries from informatician Casey Greene at the University this year. Based on the Texture editor, the tool reference managers such as Zotero or Mendeley, of Pennsylvania in Philadelphia, the tool was will allow eLife staff and authors to create and or query external databases directly. The ‘cite’ designed to manage the writing of a review work with JATS XML documents from beginning button in Authorea, for example, allows users to article on deep-learning — and coordinate its to end. Manuscripts.io can already import search PubMed or CrossRef, or pull in articles by three dozen authors. The challenge: keeping JATS-formatted content, Piipari says, and it, DOI or URL. In Fidus Writer, references can be track of which collaborators contributed together with Fidus Writer and the Stencila added from Zotero with a simple drag-and-drop. which bit of text, line by line. “We expected plug-in can export to that format as well. Manubot features cite-by-identifier, which to have a large number of contributors and Authorea allows authors to directly submit builds bibliographies using a DOI, a PubMed or we wanted to be able to look at the ‘atomic’ articles to around 41 journals and preprint arXiv identifier or a URL, without the need for changes of one person and one group of archives, according to founder Alberto a reference manager. Inserting “@doi: 10.1371/ changes,” Greene says. That is, instead of Pepe — and to embed interactive figures, journal.pcbi.1007128” into a Manubot article, navigating a tangled mess of tracked changes, executable code and data. Roberto Peverati, a for instance, instructs the tool to find and Greene wanted to be able to review each computational chemist at the Florida Institute insert a reference to the paper itself. change individually, and to keep the online of Technology in Melbourne, was asked Botvinnik calls this approach “pretty draft automatically up to date. to contribute to one such journal, Wiley’s magical”, because it circumvents the problem Manubot solves those problems by cobbling International Journal of Quantum Chemistry, of researchers using (and trying to synchronize) together various open-source tools, says Daniel in part to test drive Authorea. “I found it really different reference managers and libraries. “I Himmelstein, a Greene lab postdoc who helped very pleasant,” Peverati says. like that I can just use the DOI and it works, and to lead Manubot’s development. These include As such tools gain traction, scientific articles everyone else knows that there is one source of Pandoc, which provides file-conversion func- become ever more dynamic – and responsive. truth for the citation: the DOI,” she says. tionality, and GitHub Actions, which automates On 20 March, Greene’s postdoc researcher Authorea and Overleaf support LaTeX, the functions such as document creation. To set Halie Rando created a Manubot project to typesetting language preferred by physicists, up a Manubot project, users clone a dedicated try to make sense of the exploding COVID-19 mathematicians and computer scientists. In GitHub repository to their computer and mod- literature. Within days, dozens of researchers 2017, CERN, Europe’s particle-physics labo- ify it using a standard programming text edi- had expressed interest in contributing. “With ratory near Geneva, Switzerland, adopted tor, such as Emacs or SublimeText. Changes are something as fast-moving as COVID-19, we Overleaf as its preferred collaborative author- then pushed back to GitHub, which logs them have an urgent need for consilience, but many ing platform; some 4,800 users have signed and rebuilds the document in HTML, Word members of the scientific community are more up, says CERN computing engineer Nikos or PDF format. Collaborators can modify the isolated than usual,” Rando explains. Manubot Kasioumis. LaTeX is quite an advanced system, manuscript by submitting changes in the form provides a forum for these far-flung research- however, so Authorea and Manubot might be of a GitHub ‘pull request’ (explore our example ers to work together.