How to write a paper?
Depending on the stage of your academic career, you might either hate or love writing. Regardless, you enjoy a good story. Writing a paper is like writing a story, even though it might not seem like it.
As a college student, I used to despise writing reports and essays; I was driven instead to equations and numbers. Text just seemed verbose and inefficient, while numbers seemed rational and direct. Some of my friends were much better at blabbering through those reports. I simply couldn't be bothered.
During my Ph.D., I started learning the importance of “Writing in the Sciences,” when I enrolled in the course with the same title on coursera. I learned that journal articles (one of the foundations of academic research) did not need to be entirely dry and boring. They could be written in an active voice and used to convey a story.
As I began my postdoctoral fellowship with Dr. George Whitesides, I start understanding the value of writing as a way of expressing your thought process. In academia, if something is not published, it is as if it never happened. We have just published our first review paper from my lab (illustrated here) and about to publish our first primary paper (recently accepted). So, I figured that now is a good time to share the writing process (as previously alluded to by my colleague Dr. Engelberth).
During my postdoc, I learned that writing needs to be a part of the research process rather than simply being tagged on at the end. Using outlines is a great way to accomplish this objective. Just as I described for grant proposals, before typing up an outline, I typically sketch out a web diagram using a single piece of paper with all the major sections of the paper: introduction, results and discussion, conclusions, and materials and methods. It is useful to draw this web diagram early in the research process, for example, when you get preliminary results from your first set of experiments. For a review paper, you want to start with this web diagram if you already know the field. If you don't know the field yet, you want to perform a preliminary literature search focusing on the most recent review papers. I like using Web of Science because it allows me to filter results by article type, date, number of citations, etcetera, which is not possible when using Google Scholar and not as straightforward when using PubMed.
Within the web diagram, you want to write down what the section headings of each segment of the paper would be. A common tendency is to describe the research project chronologically; instead, you should describe your most important results first (although the materials and methods section could still be described chronologically). This approach is similar to how TV shows and movies sometimes start with the climactic moment to use as a hook and then spend the rest of the episode/movie explaining how the characters got there (e.g., see Netflix series Biohackers). When writing a primary research paper, since you wouldn't have all your results yet, you want to sketch out the types of figures that you expect to get according to your hypothesis (and be prepared to revise the expectations if you find something different).
I used to think that the introduction of the paper could simply be written at the end after all the other sections were complete, but the intro is where you set up your story and following a defined format helps set the story well. As my postdoc mentor has explained in his paper, the introduction should answer the following questions in order:
What is the problem you're solving, and what is the answer? Jump right into it with the opening sentence if possible; don't spend several sentences beating around the bush.
Background: What has been done by you and others? This section is meant to be a short literature review of the most important and relevant pieces of work. You should use the section to cite appropriate references such that when somebody is reading your list of references, the titles of first five to ten references will explain what this paper is about, even without reading any of the text. This section is useful for citing how the problem has already been tackled by others in the research community or by yourself in previously published work. This space should also be used to list the remaining shortcomings of previous approaches; ones that you will overcome in the current work.
How are you solving the problem? Next, describe the highlights of what you have done in the current work and its significance.
Listing the above points on the web diagram help keep a cohesive story on a single piece of paper. We have a hard time remembering multiple ideas in our head simultaneously and thus having them listed in front of us is beneficial.
After sorting out the introduction section, you can focus on the results and discussion. Within the results section, you want to sketch out your most important results (or expected results) in the main text and decide what figures or tables you would use to convey these results. Designing figures and tables will take considerable time because you want to convey your message clearly (instead of just using these illustrations as a means of dumping data). There is an excellent series of articles called Points of View published in Nature Methods, which is a useful guide to consider when drafting figures for data visualization. The figures that convey your story should be used as the main figures, while all others should be presented in supporting information or appendices.
The discussion section (often combined with results) is where you would compare your work to that of others and provide the implications of your findings. Thus while the results section is mostly descriptive, the discussion section involves interpretation and comparison.
Building from the results and discussion is the conclusion section. Here you can list (I mean actually list in a numbered format) out the key advantages of your work compared to previous works. It is also a good place to list a few limitations that are yet to be overcome. Finally, you want to end on the note of the potential for this work in future studies or applications.
The materials and methods section is typically the easiest to write for any scientist/engineer because it is the most technical and it is describing the work that was completed. The key here is to provide lots of detail. With increasing scrutiny on reproducibility, this section needs to have all the details that would be required the reader to reproduce the experiments and findings. Since journals will often have word limits on the main text, supporting information or appendices are a good place to describe all the methods in detail.
Once you have drafted an outline, you can expect around 10 to 15 rounds of revision with your co-authors, before you are ready to submit the manuscript to a journal for peer review. One submitted, the manuscript will of course be scrutinized independently and will often require further revisions before it is finally accepted for publication. The whole process could take around 1-2 years of work and $100-200k in spending per paper. These papers together build a body of knowledge and advance science and technology.
In academia, peer-reviewed publications also serve as a metric of productivity. It is important to publish frequently and in good quality journals. Your authorship is public and your peers will use your Google Scholar page to determine what your most recent findings are. As a tenure-track faculty, my publications are an important component of my evaluation for tenure. As a student or postdoc, publications are used for evaluating your performance for a job (whether in industry or in academia as I described earlier). The more you practice writing (and revising), the better you will get at it, and the less you will despise it. So, start writing!