The Grand Locus / Life for statistical sciences



Journal clubs, ranked from worst to best

One of the most difficult tasks for an academic is to know the literature. Most labs run some kind of literature discussion, which are usually referred to as “journal clubs” in biology. We tested several variants when I created my lab in 2012 and we have learned a good deal about what works and what does not. So here is my personal view on the different types of journal clubs, from worst to best.

5. Online discussion forum

Live meetings have several disadvantages: they disrupt the workflow of experiments, there are no records, and some people speak too much. I reasoned that a good way of addressing all this would be to put up an online forum where we could upload a paper and share our thoughts. I invited all the ~300 researchers of the institute to participate with the plan to go to social media if it gained momentum.

That did not work at all. We never got past the first paper, which had only one comment (mine). It was not appealing at all to formalize your thoughts in order to write a paragraph and post it for people who work next door. The written form is unforgiving and brings a slew of misunderstandings, taking away the lightness and the joy of communicating with each other.

Back then in 2012, we did not yet have an overdose of online interactions, and still, the proposal was very strongly rejected. Maybe this could work in other contexts, like for a team over multiple sites and with a lot of screen time. Otherwise, the appeal of live discussions is too strong to resist.

4. Powerpoint presentations

This seems to be the most common type of journal club in biology. Someone reads the paper, prepares a Powerpoint presentation, shows the figures on the slides and guides the audience through the results. I never tried this type in my laboratory but I had the chance to see it in action when I was a PhD student.

One of the main issues with this type of literature discussion is engagement. Most of the time, the presenter is the only person who reads the article. The audience processes the information at surface level, without the depth that reading can provide. Also, the audience often turns into an angry mob and aggressively lynches the work. I never understood why and I wonder if anyone in the community has an explanation... Fortunately, I have also seen structured discussions, especially if senior scientists enforce it.

On the flip side, this model of discussion is rather good for debates or to propose new ideas, as if the lack of engagement gave more freedom to sidetrack.

The reason I did not try this method in my laboratory is that I believe that students should read a lot, because reading scientific articles is an important skill to develop. They cannot acquire it by attending presentations; for this they have to read upward of 100 articles.

3. One figure each

In this model, everybody has to read the paper of the week because the presenters are selected at random. Each person describes a figure, starting from the first. This version is low maintenance: everyone comments or follows from their copy of the article; there is otherwise no material or particular preparation.

I found the engagement to be high in this variant — there is a social cost to showing up without reading the article — which helps build reading skills. The feeling of shared responsibility to describe the article also sets a good tone for discussion or debate, so this method strikes a very good overall balance.

The one thing I have against it is that the quality of the discussion often goes down toward the end of the article. Many participants read the beginning of the article but drop halfway. Only few participants can comment the end of the manuscript so the discussion often turns into a monologue. This is probably a limitation of the model of shared responsibility: participants do commit, but not fully.

2. One paper each — freestyle

In this variant, each participant describes an article of their choice. Since there are multiple articles per session, they do not have enough time to go as deep as the single-article variants.

The main limitation of this method is obvious: descriptions and discussions can be shallow. One may even say that it does not promote critical reading, but this is not my opinion. At some point in their PhD, everybody realizes that some papers are crackpot. But not everybody develops a deep understanding of their research field...

This method shines by its coverage. Summarizing an article is an excellent exercise to develop efficient reading skills. At the same time, the increased bandwidth of going through multiple articles per session should allow students to cover the basic knowledge of their field faster than with other methods.

1. One paper each — coordinated

This variant is almost the same as the previous one. The difference is that all the articles of the session are chosen by the same person. This is my personal favorite.

The disadvantages are the same as for the previous variant. Coordination essentially brings two benefits. The first is that the coordinator is forced to choose a topic, screen several articles and make a selection. This helps develop a sense of relevance and allows students to improve their literature search skills. The second benefit is that it brings landmark articles into focus. To discuss certain recent discoveries in a given topic, it is often needed to go back to the roots and introduce older ones, which is only possible if all the articles of the session are connected.

We practiced this method with four articles in one-hour sessions, allowing us to go through over 100 articles per year. 15 minutes is extremely short to discuss a scientific article. Sometimes it was too short, but we always learned a lot and read many good articles.

The one disadvantage that I see is that this method is very hard for junior scientists. Reading an article is hard, let alone summarizing it in a few minutes. It can be frustrating for them and for the other participants if they are unable to keep up the pace. Most beginners will be grateful to go through the article with their supervisor at least a day before the live session.


Literature discussions serve more than one purpose. What works for some labs is maybe not what works for your lab. I consider that the most important is to develop scholarship and knowledge of the field. What about you? Which skills do you think literature discussions should build?

« | »

blog comments powered by Disqus

the Blog
Best of

the Lab
The team
Research lines

Blog roll
Simply Stats
Ivory Idyll
Bits of DNA