Writing a scientific paper
This is the Netoff lab approach to writing a paper. I like to do things in a particular order, stemming from the data and then writing the text.
Determine the audience
- What journal will this be published in?
- Formatting of the journal for text, figures and references
- What will be basic knowledge presumed by the readership be?
- Will you be publishing color figures or B/W?
- Make sure all axes are labeled. Set font sizes so that they are readable when figure is in published size.
- Title all the figures
- Publication quality figures can be
- Vector graphics, such as eps files, this is best for line graphics
- Tiff files with 600dpi or greater, this is best for images
- Often a journal will describe the accepted formats, check before you get started
- Size all of your figures for final publication. If the journal uses columns, they will provide widths for 1c, 1.5c and 2 column widths. Otherwise, choose half or full page widths and set heights to have appropriate aspect ratios.
- Captions should describe necessary details to understand the figure, should be stand alone
- Place figure in a textbox in word.
- Right click on figure and add caption to generate a link that automatically numbers the figure. Then, write the caption.
- In text use crossreference to insert link to figure.
General advice for making figures:
- The less ink used to convey the same information the better.
- Get rid of bounding boxes and extra ticks.
- You can use scale bars instead of axes often and eliminate a lot of ink.
- Use black lines and dots unless you plan to use color in the figure
- When using color images that may be printed on a B/W printer, think about how the colors will convert to B/W. Try to use both line styles and colors so that they can be distinguished even if the figure is not in B/W.
- Be concise, stick to just the details needed to reproduce your findings
- If you are using models or techniques described in other papers, give at least a short description of them in your paper and then cite the source
- Make sure to include all concentrations and units
- Describe all statistical tools used
- you can leave out details such as the kind of integrator used, if talking about a computational model, as long as you make it clear that you addressed important things like determining an accurate time step
- Stick to describing the actual findings
- Provide all the statistical results
- Describe what the figure is about in the body of the results and crossreference the figure, do not provide details that are covered in the caption
- You must have a link to every figure in the text
- Do not include discussion points, such as interpreting the data, unless it is necessary to describe why you did the next set of experiments. Do not compare your findings with other here. All this belongs in the discussion section
- You should point out any deficiencies in the experiments and interpretation of findings here (this is good rhetoric, rather than letting the reviewer point them out)
- You can outline what experiments might be done in the future
- This should start with a broad picture: What is the general problem and why do you want to do this work. For example, "Epilepsy is bad and we want to prevent seizures" not "We want to measure PRCs from CA1Pyarmaidal neurons using a new method"
- This should put your work in the context of what other has already been done in the field and why your work is needed
- It should then give a brief outline of the central hypothesis to be tested and how it will be done to introduce the methods and results section. Don't write about conclusions here.
- This often can start with a brief overview of the paper
- List the major findings and their implications
- This section should be short and sweet
Write your abstract at the very end. There are often word limits to the abstract. It should contain:
- General statement of the problem
- How you propose to solve it
- Methods used
- One or two sentence about the findings
Try to find a concise title that describes what your paper is about. It might be good to get some outside opinions. If people don't understand what it is about from the title, they are not likely to read the paper.
It is best to use a reference database, such as EndNoteWeb. This is free through the university EndNoteWeb
Install the citation software on your computer
Go to pubmed and select citations
Insert citations from your database into your paper
Format the references selecting your journal's style. You may need to generate a style for the journal if it is not in the style database, or modify from another journal.