Reading Primary Scientific Literature

There are different ways of communicating science: Dance Your Thesis

Perhaps the most direct are primary research articles.

Elements of a scientific paper

A workflow for reading papers

Breaking down the steps of reading primary scientitif literature

From Jennifer Raff's blog post: How to Read and Understand a Scientific Paper

Working through the workflow..

Working through a paper

A general bio data science paper: A saturated map of common genetic vairants associated with human height

Step 1. Read the INTRODUCTION

• have to skip over the abstract even though it is the first thing presented and (sometimes) the only free section
• write down every word you do not know or is confusing in its context

Step 2. Identify the BIG question

• circle sentences from the introduction
• succinctly re-write what question this paper is trying to answer or problem is it trying to solve
• focus on why the research is being done
• be wary of agenda-motivated research

Step 3. Summarize in the background in < 6 sentences

• what work has been done in the field already?
• what knowledge gaps or limitations remain?
• according to the authors, what should be done next to address these?

Step 4. Identify the SPECIFIC question(s)

• what question exactly are the authors trying to answer?
• are the multiple questions or just one?
• identify any null hypotheses being tested

Step 5. Identify the approach

• what are the authors going to do to answer the specific questions?

Step 6. Read the METHODS

• draw a diagram for each experiment
• write down methods you need to look-up to understand the results
• if it is a "methods paper", look for context with notes from the introduction
• note if any methods reference supplemental material (and where the main methods are...)

Step 7. Read the RESULTS

• write some paragraphs to summarize each paragraph, figure, table
• don't worry about what the results mean yet, just note what they are
• read figure and table legends carefully - the information in them may not be in the body of the text
• pay attention to the words "significant" and "non-significant" (stats)
• do graphs have error bars? Do they show the original data and sample size?
• check supplemental data, available at the end of the paper or online

Step 8. Determine whether the results answer the specific questions

• now re-read your notes and think about what the results may mean
• form an opinion before moving on - it is OK to change your opinion later
• note if any results are unexpected given what you noted in the introduction

Step 9. Read the DISCUSSION

• what do they authors think/say what the results mean?
• it is OK if your opnion is different
• can you think of any alternative explanations for the results?
• are any weakness or limitations noted? can you think of any (others)?
• what do the authors suggest as next steps and can you think of different ones?

Step 10. Read the ABSTRACT

• does it fit the paper as you now understand it?

Step 11. Explore what other researchers say about the paper

• try and find experts in the fields and if they have commented
• are there any criticisims you had thought of, or new ones?
• check in any general publications have picked up the story
• find other papers that cite this one and what they say it says
• are the public reviews of the paper or its supporting grants

Open peer review: Peer review file

Step 12 cont.

A news release: Largest genome-wide association study ever uncovers nearly all genetic variants linked to height

Step 12. Take a break, then repeat as necessary

• this is a lot if information to take in, give it some time to consolidate
• experts in any field take multiple passesses to fully understand a paper
• on subsequent reads, re-use and mark-up your notes to build on them
• data science papers often have (funding-mandated) data and code availability links

In Summary

Further resources

• Annotated papers: Science in the Classroom