13 July 2014

I had to read 30 papers on ebola pathogenesis this week. When I was digging my way through paper number 7, I made a flawed observation: reading papers is boring. Most graduate students would agree with that statement, but I want to convince you that calling papers boring is as pointless as saying rocks are stupid. Boredom comes from the mismatch between the way we would like to consume information and the way it is presented. Understanding this difference can make our presentations more interesting and our stories more engaging.

Look at Figure 3 from Bradfute2010 (it shows two different spleen sections treated with a stain that darkens apoptotic cells):


Figure 3: Overexpression of Bcl-2 protects lymphocytes from EBOV-induced apoptosis.

The Results section describes this figure by saying:

Vav-bcl–2 mice showed nearly complete protection against lymphocyte apoptosis compared with wild-type littermate control mice (Fig. 3).

That’s a fact.

It’s not a story yet because it doesn’t provide us with context to fill our knowledge gap. For example, what are Vav-bcl-2 mice? The Materials and Methods section briefly mentions them:

Vav-bcl–2 transgenic mice, in which Bcl-2 is overexpressed on all cells of hematopoietic origin (20), were a kind gift from Dr. Jerry Adams (Walter and Elisa Hall Institute, Melbourne, Australia).

You might be wondering why anyone would want to overexpress Bcl-2. You can find a clue in the Introduction:

The intrinsic [apoptosis] pathway can be inhibited by overexpression of Bcl-2, which stabilizes the mitochondrial membrane and prevents the release of cytochrome c.

If you’re still not sure how to interpret this information, you can find the authors’ perspective in the Discussion:

[…] it was surprising to find that transgenic overexpression of bcl-2 in all hematopoietic cells conferred near complete protection from EBOV-induced lymphocyte apoptosis. There are at least three explanations for the protection from EBOV-induced lymphocyte apoptosis in vav-bcl2 mice. […]

If you combine and expand those four sections with many additional details, you might have a story. It’s kind of weird, but we currently write scientific articles by choosing a few experimental results, splitting them into parts (background, findings, interpretation and technical details), and grouping them into longer sections (Introduction, Results, Discussion and Methods). Then, the readers have to read the entire paper and combine each individual story in their heads. I would love to change the way we write scientific papers to make them easier to read, but that’s not the battle I’m picking today. Instead, I want to focus on how to become a better storyteller by reading papers more efficiently.

Read on a need-to-know basis

When you’re just starting to read papers, the most common mistake is to think that they can be read casually. You start reading, you stumble over a pile of confusing facts, and you get dragged away by boredom and sleep. Instead of poring through everything between the abstract and the conclusion, try reading on a need-to-know basis:

  • Identify the pieces that make up the title (and simplify them). It’s easy to miss the meaning of strung-together words unless you dedicate some time to unpack them.
  • List all the research findings mentioned in the abstract (these are useful to determine if the paper is relevant for your research).
  • Skip the Introduction and study the first figure.
  • If you understand it, move on to the next figure. If you don’t, find the paragraph in the Results section that describes it.
  • If you don’t understand something in the paragraph, go back and look it up in the Introduction (or if it points to a reference, read that paper before you continue with the current one).
  • Once you’ve understood all the figures, skim the Discussion for additional insights.
  • Write a summary of everything you found interesting (this is key).

Need-to-know reading is a better way of consuming scientific information than traditional reading because it forces you to identify and fix gaps in your understanding before you move on to new ideas. It helps to have the bibliography next to you while you read (most PDF viewers have a “split screen” function, so you can look at two sections of the same document at the same time) and to check out the original references every time you run across something you don’t understand. I like to build a map that shows the path I’m following between papers—it prevents me from getting lost and it helps me decide what I should read next (it also saves me from having to open 40 tabs on my browser).

Reference Tree A section of this week’s literature map

You don’t know something unless you can summarize it

The most important part of reading a paper is the summary you write about it. This is part of the one I wrote for Bradfute2010:

  • The Bcl2 protein prevents mitochondria from releasing cytochrome c and causing apoptosis.
  • The authors show that mice overexpressing the Bcl2 gene don’t undergo lymphocyte apoptosis during Ebola virus infection.
  • The authors think that three likely explanations are X, Y and Z.

I’m pretty sure that a month from now the only thing I’ll remember from the 30 papers I read this week will be the summaries. It takes discipline to come up with a summary for each one, but it also makes it easier to connect multiple pieces of research (Bradfute2010 says that Bcl2 was important but Morgan2020 contradicts it), and to discuss my conclusions with other colleagues. Ultimately, the better you get at translating facts into stories, the better you get at presenting.

If you’re not reading as many papers as you’d like to, or if you find the experience frustrating, read your next paper on a need-to-know basis, build a map with the references to other articles (so you can decide which ones to read next), and write down a summary of the things you find interesting about each one. Reading something is only useful if you can remember it.