Tips for writing your first scientific literature review article

When I undertook the task of writing a scientific literature review article last year, I had hoped that a Google search would reveal a handful of how-to pages thoughtfully created by veterans of this particular writing process. I found nothing of the sort, so I plowed ahead on my own, inventing techniques for myself. I’m now offering this piece for other young scientists who find themselves in similar situations. What you’re reading now is basically a case story with an N of one, but it is the sort of essay I wish had been available to me when I started.

I was running a protein over a nickel column on a Sunday evening in February 2010 when my adviser approached me about co-authoring a review article for Annual Review of Biochemistry. My adviser is a busy guy, with a lot of papers and grants to work on, so I knew that by “co-author” he meant that I would be the main researcher and writer, getting mostly broad, guiding suggestions from him. That was fine with me – as a fifth-year graduate student, I had learned to cope with, and even prefer, extreme independence. To be honest, I was excited to have this opportunity to examine the literature in depth and to create something useful out of it. The due date was August, so I had six months to synthesize decades’ worth of research papers on our topic into one conveniently sized, nicely packaged bundle of facts and interpretations

Getting started

Our topic was caspase substrates, a diverse group of proteins essential for programmed cell death and thus important to our understanding of how to kill cancer cells. A PubMed search for “caspase substrates” yielded more than 2,000 research papers. I had no illusion that this project could approach comprehensiveness, and luckily my adviser didn’t either. I would have to assess the limits imposed by the journal (30 pages, six months) as well as my own limits and the necessity to balance the writing project with lab work that was essential to finishing my Ph.D.

Narrowing the scope of the article to conform to these boundaries was perhaps the biggest challenge of this process.

Knowing that I work better when I focus on one project at a time, I spent the next two months carrying out all of my regular lab work while only pondering the review article and skimming the literature when I had time. After that, I transitioned to full-time reading and writing. I found a café that I liked in my neighborhood and spent nearly every morning there that summer drinking tea, eating pumpkin muffins and working on my laptop. Afternoons I often spent writing at my apartment or at the library on campus. I knew that concentrating on the article in my crowded, noisy laboratory would be impossible, but it also was essential to spend some time there each week consulting with my labmates on my literature research, keeping up with lab business and gossip, and retrieving my ergonomic pipettes from other peoples’ benches around the lab (they always seemed to get kidnapped as soon as I posted a “working from home” status update on facebook

The finished product

There were many points at which I felt overwhelmed by the task and didn’t see a clear path to finishing the article on time. I tried to reassure myself by remembering that I had been rather good at writing term papers in college; but this was a larger task and one with the potential for having an impact on someone, somewhere, sometime who wanted to learn about caspase substrates. In the end, I finished by the deadline (well, plus one two-week extension the editor agreed to grant me) and was very happy with the product and with all I had learned about caspase substrates, about the scientific literature and about the review-writing process. Yet I estimate that the next time I undertake a task like this, I’ll be able to do it in half the time. I hope the following tips will help other scientists who find themselves in this kind of uncharted territory.

I’ll end by mentioning that, for me, this was one of the most rewarding experiences I’ve had during my time as a Ph.D. student. Distilling all sorts of data from experiments done by scientists all around the world into a coherent story turned out to be very satisfying. I look forward to doing it again someday, perhaps in a somewhat more efficient manner.

Define the scope of the article. Make an outline, keep lists of topics that are and are not within your scope, and remind yourself to stop any time your reading wanders outside your scope. My adviser and I settled on devoting the first half of our article to a broad survey of a few key research topics (for example, the physical details of the caspase-substrate interaction) and devoting the second half to a few highly detailed vignettes about some of the hundreds of known caspase substrates.
Your labmates and collaborators are invaluable resources. Each has a specific area of expertise that’s probably slightly different from your own. Ask colleagues which papers they’d give to a rotation student to read and what the most important recent advances are in the field. (Be careful not to let this lead you too far astray. Your colleagues’ ideas may help you define your scope when you are starting out, but you do not have to incorporate all of their suggestions if you don’t feel they’re relevant.)
Don’t dwell on previous review articles that have been written on your topic (this quickly can become a black hole that sucks up time and gives you unnecessary insecurity about the contribution you’re trying to make to the field), but do familiarize yourself with their content. Look for areas that have not yet been thoroughly reviewed or areas for which you think you have a fresh take on old data. One of the most painful things that can happen is to spend days reading and writing about a topic only to notice later that there’s a section of another review article that explores the same area, references the same set of papers and comes to the same conclusions.
Make yourself comfortable. This may seem obvious, but I think it’s important. Find places to write where you can concentrate, and take breaks often to stretch, get a snack or even step outside for a few minutes. On days when I struggled with concentration, I often used a timer to structure my day. I would work for 60 minutes, then take a sanity break, then work for another 60 minutes, and on and on.
Impose some structure on the mess that is the scientific literature. I developed a strategy for each research topic that I wanted to review (including the broad survey section in the first half and the vignette sections in the second half). First, I found the most recent papers on the topic and went through them, picking out what looked like important references. I worked my way backward to a set of about 10 key papers. Then I quickly read and made a summary for each, usually in the form of a bulleted list of the conclusions drawn from each figure. Next, I combined those summaries into a single table. (I did this by hand on paper; an Excel spreadsheet also would work). Each research article was one row (arranged by publication date), and the columns were results or conclusions reached. I then easily could see which papers agreed on which topics, what trends emerged over time and where the controversies in the field lay. I found that once I had made a table, the narrative of that particular research topic almost wrote itself.
Spend some time writing with all your PDFs and Web browsers closed and your desk cleared of any paper. This was advice my adviser gave me about a month before the due date, when he could tell that my brain and my PDF library were so overflowing with data that I was struggling with actually producing any text. I didn’t find it easy at first. I didn’t want to get anything wrong, even in a draft, so I was afraid of typing even a single sentence without references to back me up. On the other hand, with the Internet and all my PDFs in front of me, I tended to generate sentences that were very dense with information but not necessarily closely related to each other – and not always pertinent to the specific scientific narratives I was attempting to compose. I started making real progress on the writing only when I spent a few August afternoons sitting on the roof deck of my apartment building with a pen and paper and no Internet-capable devices. Yes, I sometimes wrote things that were wrong (or at least imperfect) when constructing a section from memory. However, I often ended up with a strong scaffolding onto which I could later add some of those dense, fact-laden sentences.
Don’t be shy about clearly defining your role relative to that of your co-author(s) before you begin, or even along the way, if you feel amendments are needed. This was easy in my case, because my adviser and I both preferred that I be the main researcher and writer and that he act as a consultant on high-level issues. However, I am keenly aware of other cases that did not work out nearly as congenially.
Read the journal’s instructions for submissions carefully. You should have the email address of an editor at the journal; don’t be shy about asking questions. Do not ignore the journal’s page limits or formatting requirements. Pay very close attention to the graphical requirements for figures. Make sure to get permission to reproduce any figures in your review. (This usually is done by following the permissions instructions on the website of the journal in which the original figure appeared. It’s also not a bad idea to email the authors who made the figures to let them know that you will be using their work).
Get familiar with software like Papers (or any other PDF-management software), EndNote and Adobe Illustrator (or whatever graphics program the journal suggests). For me, online Adobe Illustrator tutorials provided nice breaks when I’d been reading for hours and hours.
Your labmates and collaborators also can help you with the editing process. Rather than asking one or two people to help you edit the entire article, break it up into sections and ask a different colleague for his or her expert help in revising just one section on a topic with which you know he or she is familiar. Another strategy is to give part or all of your article to a first-year graduate student or to a scientist in a slightly different field. He or she is your target audience and will let you know if there are sections that need to be revised