Thursday, 26 March 2015

Glycoinformatics, I CHOose you!

Next meeting: Tuesday March 31st

This week at Journal Club we are going to branch* out into area of glycoinformatics (or glycobioinformatics) -  that is the bioinformatics of carbohydrates (CHOs).


Lactose is a carbohydrate.


It's an area that is not commonly part of most introductions to bioinformatics, but exposure to them is literally in our blood*.

Blood type antigens

The first paper provides some background and outlines the state and challenges of glycoinformatics, while the second deals with the online database for glycan structure, UniCarbKB.

Along the way, we'll touch on some database sustainability issues that have been addressed in UniCarbKB, and see what shape* the nomenclature for glycan structures is in.


Consortium for Functional Glycomics (CFG) Nomenclature (http://www.functionalglycomics.org/)


The two papers are:


I'm sure the discussion will be sweet*.






* All puns are intentional. I apologise for nothing.


Wednesday, 21 January 2015

Document Preparation and Its Discontents

To kick off 2015, the theme of yesterday's Journal Club was about increasing academic productivity. Two blog posts dealt with effective annual goal setting and "New Year's resolutions for scientists". Although interesting and full of useful ideas, these were rather uncontentious.

In contrast, some rather impassioned discussion was – perhaps predictably – inspired by an article in PLOS ONE with the understated title of An Efficiency Comparison of Document Preparation Systems Used in Academic Research and Development.

From the moment it was first published about a month ago, this article has been generating a lot of heated internet discussion. Both the comments section and the media coverage page reveal widespread indignation and scoffing. The authors have also made an extensive comment in which they answer some of the objections. (I'll keep referring to this as the response.)

Someone is wrong on the
internet

The discussion around our Journal Club breakfast table went on to mirror the same debate. Fortunately we're all friends so we managed to avoid taking it out onto the streets in full blown internecine warfare; although, it was close!

Concisely, Knauff & Nejasmic set up a study comparing LaTeX to Microsoft Word and their conclusion favours the latter for most tasks. The article has been criticised because of biases in the study design, and because its discussion takes results from this study and escalates to some strong assertions. Effectively, the authors proscribe the use of LaTeX except in certain circumstances; furthermore, they impugn the judgement of LaTeX users, not only in the article per se but again in their response to comments.

Since these criticisms can be found by going to the comments and media coverage sections linked above, there'll be little need to repeat them here. The intent here will be to focus on something that doesn't seem to have been targeted by many other comments or blog posts. This is, that the authors clearly feel entitled by their psychological training to make assumptions about their opponents' intellectual faculties. They do this without referring to any criteria of evidence. Their basis for measuring cognitive impairment is whether or not people agree with their conclusions.

Both article and response ring with cries of "cognitive dissonance"; and whereas the article attempts to justify its proscriptions by invoking "the taxpayer", that eternal hired thug of anti-intellectualism, the response more clearly reveals the authors' biases against LaTeX. The response is quite telling in the way the authors offer a blanket psychological assessment gratis to people who enjoy using LaTeX.

Namely, if you think it's important to keep content and structure separate to formatting instructions, then you are suffering under some colossal convergence not only of "cognitive dissonance" but now also "signalling" and "polarisation", with a bit of confirmation bias thrown in, oh my. This cutting analysis is mainly focused at LaTeX users, but they don't hesitate to advertise how they "think" that markdown "probably" shares many of the same problems.

Now, the authors belong to the Department of Psychology, Experimental Psychology and Cognitive Science at the University of Giessen. We can suppose, then, that they are psychologists. Maybe I'm suffering a bit of cognitive dissonance here, but I'm under the impression that psychologists would not consider themselves in a position to make sweeping statements about a diverse group of people without first conducting a study to investigate that group.

Note that this wouldn't be the same study as the one they've already undertaken, which was (ostensibly) concerned only with productivity differences between LaTeX and Word. In order to make broad assertions that LaTeX users suffer from cognitive dissonance and confirmation bias, reinforced by signalling and polarisation, then a study would need to be conducted on LaTeX users to see whether this characterisation has any basis in reality whatsoever.

Furthermore, by summoning forth all these psychological demons to support their overstretched assertions, do they not risk hoisting themselves by their own petards? For making guesses about other people's cognitive dissonances, confirmation biases, signalling and polarisation is a saloon door that can hit someone on the rear both while entering and leaving premises of ill repute.

After all, it's quite clear that Knauff & Nejasmic are indeed suffering, you guessed it, cognitive dissonance, confirmation bias, signalling and polarisation.

Cognitive dissonance: there is a disparity between the authors' scientific training and their jumping to conclusions based on biased study design. When this is brought to their attention, they attempt to reduce dissonance with unsatisfactory justifications (i.e. intensifying their poor arguments about LaTeX users' poor cognitive abilities), and by ignoring conflicting information ("Do we compare apples and oranges? We think that is just verbal hairsplitting ...").

Confirmation bias: this has been dealt with by those who have criticised the study's design, and the jump from the numerical results to wide-reaching conclusions. Furthermore, the response concludes with weasel words that strongly indicate their pre-existing bias: "in our group nobody uses LaTeX anymore since an efficient way of collaborating in the same document is more important to us than individual software preferences."

In other words, their group had already reached the conclusion that they don't like LaTeX. And if they began with a mixed group that became anti-LaTeX only after running the study, it seems unlikely that they would have failed to mention this. This implies that they had an anti-LaTeX group all along. So, the study concluding that LaTeX is usually inferior was designed and run by a group that had already decided without a study that LaTeX is inferior. Bias confirmed: check.

Signalling: if anything, this is the weakest argument against LaTeX. Humans signal all the time; it's part of being a social organism. If people in STEM didn't use LaTeX to signal, we'd still have the many other cool things which separates us from lesser mortals, such as version controlling everything and using the Unix command line for everything. I don't have a problem with the idea that LaTeX is a form of signalling because the signalling arises from its usefulness to a certain kind of person. My kind of person.

But if we want to use signalling as a basis for discrediting people, then it seems a little disingenuous of Knauff & Nejasmic to pretend that they're not also sending out signals. What better way to signal your cool contempt for a vast number of uptight math/computer nerds than publish an article calling for the banning of a beloved and venerable tool such as LaTeX? And should people on both sides of the debate stop publishing articles because this is a way to signal status in academia?

Polarisation: the thing with polarisation is that it needs two polar opposites to play off against each other. In psychology, polarisation is driven by a mechanism called the "pendulum effect". This requires two groups of people to make increasingly antagonistic statements towards each other. It doesn't work with just one group who have no one to argue against.

The only way polarisation could be responsible for generating a vocal LaTeX community is if LaTeX users have found themselves in arguments against people who are anti-LaTeX. That is, Knauff & Nejasmic, and their group, being anti-LaTeX, are in fact as much the victims of polarisation as their pro-LaTeX critics. What's more, if they take polarisation seriously, then it would behove them to take responsibility for driving the pendulum ever harder with their anti-LaTeX biases and extreme conclusions.

Overall, there is a twin problem with this psychological line of argument that Knauff & Nejasmic seem committed to. First of all, if the same argument can be used to win a debate in either direction, then you should probably be cautious about relying on it.

Secondly, and I think more importantly, it seems unethical for psychologists to carelessly diagnose all comers as having cognitive defects, with no burden of proof required, just for the purpose of point-scoring in an argument. I would hope at the very least that Knauff & Nejasmic will concede that this kind of behaviour has been inappropriate. They have an opportunity now to demonstrate their graciousness by apologising to LaTeX users for the numerous insults, and retracting their offensive and ill-conceived judgements.

Well, since we seem to be operating in a post-scientific wasteland where unsupported assertions are fine so long as no taxpayers are having their minds stretched, it must be safe to conclude with this xkcd which was not created using any taxpayer money:

XKCD: Trustworthiness of File Extensions

NB: if you found all this annoying and your primate brain chemicals are urging you to flame someone, you may cross swords with me on twitter.

Saturday, 11 October 2014

Ten Simple Rules for a Bioinformatics Journal Club

We are pleased to be presenting a poster based on our experiences creating and running this journal club at the first Australian Bioinformatics Conference:



Full size PDF (A0) available here

Friday, 7 March 2014

Publish and perish? Papers, blogs, criticism and the internet.


In recent meetings, journal club members have selected blog posts alongside traditional peer reviewed papers. These often complement the topics, and can offer practical, informal supplementary material to the published papers. 

The contrast between them has led to some discussion on the place of blogs in science. Journal clubs members are adept at the technical platforms behind them; most of the journal club is organised through a complex combination of email, Facebook, mailing lists, Google Drive and determination. Many maintain their own blogs and have active social media presences, but there has been a sense of not wanting to waste material on a blog post. My personal opinion is that the more 'runs on the board', that is, the longer your lists of published papers record, the more likely you are to feel confident in blogging the work (or preview of the work) rather than publishing.

The differences between published papers and blog posts, criticism of each and the role the internet plays in all of this will be the topic this week. Note that the goal this week will not necessarily be to discuss the merits of all papers in depth. Rather, I'd like to discuss the issues raised by the following papers, blogs and social media commentary, and determine if it is possible to publish and perish, and if so, how to avoid it. 

"Quilt plot"
Paper: "Quilt Plots: A Simple Tool for the Visualisation of Large Epidemiological Data" 
Response: PLoS one comments  and a blog post with commentary on the above and relevant links to other reactions.

Networks
Papers: 
Network link prediction by global silencing of indirect correlations
Network deconvolution as a general method to distinguish direct dependencies in networks

Response:
http://liorpachter.wordpress.com/2014/02/10/the-network-nonsense-of-albert-laszlo-barabasi/
http://liorpachter.wordpress.com/2014/02/11/the-network-nonsense-of-manolis-kellis/
http://liorpachter.wordpress.com/2014/02/12/why-i-read-the-network-nonsense-papers/
http://liorpachter.wordpress.com/2014/02/18/number-deconvolution/

Rebuttal:
http://liorpachter.files.wordpress.com/2014/02/response_to_nonsense_blog_post.pdf

Wednesday, 12 February 2014

Writing Gym

Not too long ago, I was discussing that well-known academic law of "publish or perish" with a seasoned bioinformatician. I was bemoaning my feelings of poor progress in my PhD, and pointed out that I often don't feel like I have results ready to write up.

My friend (who is real---this isn't Rich Dad, Poor Dad) observed that publications are in fact not something we have any control over. Once a piece of writing lands in the lap of some publisher, all we know is it might be accepted with some probability. It is pretty much out of our our hands as of that moment. And some pretty bad writing gets accepted sometimes, so you don't need to be a perfectionist with your submissions. If the reviewers have a problem, they'll let you know, so why not just unleash your crappy writing on them and hope for the best?

Although we don't have control over publishers and reviewers, we do have control over how much we write. So our publication rate can be maximised, on average, just by writing as often as possible. My friend even opined that the writing doesn't really need to be about our specialisation all the time. So long as it's scientific and it wouldn't be too embarrassing to have it seen by strangers, that will do for starters.

This is similar to the sentiment expressed by Stephen King in On Writing: "If you want to be a writer, you must do two things above all others: read a lot and write a lot. There's no way around these two things that I'm aware of, no shortcut." And I wouldn't want to disagree with Stephen King. He might respond by writing a story about my laptop getting possessed and eating my hands. Then it would really happen, because that is one of Stephen King's powers. And then I wouldn't be able to write anymore because I wouldn't have any hands.

Anyway, at first I was kind of skeptical about whether it was really possible for mere mortals such as myself to write all that much. I took my friend's words as encouragement to kind of think about writing more, but I hoped I could squirm my way out of it. I questioned whether it would be optimal to write for an hour a day, or a day per week, or a week per month. Probably more often than a month per year. Probably.

But at about the same time I encountered two suggestive links related to the same ideas. The first was a summary of research by Robert Boice on the Tomorrow's Professor Mailing List which suggests that academics who commit to writing a little bit every day end up being a lot more prolific than researchers who "binge write". The other was a Thesis Whisperer post describing the notion of "Shut up and Write!" meetings.

With all this churning in my mind, it seemed natural to propose that we segue the BGSA Journal Club with "Shut up and Write!"-style pomodoro sessions. Well, the traditional pomodoro technique consists of multiple sessions of different lengths broken by rest-breaks, but it currently seems that a single half-hour session is best for us. The name "Writing Gym" only came later, after running our first couple of pomodoro experiments. This wasn't really intended to distinguish ourselves from "Shut up and Write!" groups. It just seemed that "Writing Gym" is a more succinct and accurate description.

The goal of Writing Gym is to help everyone both with their short-term writing goals and also to acquire habits that will lead to more prolific research careers in the long-term. Writing in a group helps with motivation (hence the word "gym"), and we can also share ideas about what sorts of things to write and how to become better writers.

There is no expectation that writing at Writing Gym should be bioinformatics specific. It probably doesn't even have to be about science. There is no expectation that anyone is writing towards some specific deadline, although you can definitely use Writing Gym for that. We encourage writing that could potentially be shared with others, therefore no emails or facebook updates. (But, you know, it's not like we're going to get angry and kick people out if they decide to quietly procrastinate.) Examples of writing done so far have been preparing blog posts, writing conference abstracts and starting work on a submission for the Taming of Chance story competition. If you get stuck for ideas, you could always summarise one of the papers from Journal Club.

Having now had a couple of sessions, it seems like people find them useful and fun, so it is likely that Writing Gym will become an ongoing activity with BGSA. Every second week we have Journal Club, and in the off weeks, Writing Gym. The format and atmosphere (and people) are mostly the same between the two.

Like Journal Club, Writing Gym is an informal breakfast event. First, we chow down and drink coffee and chit-chat for a while. When everyone has arrived and more or less finished stuffing their face, we start by going around the table and nominating our intended writing topics. Then we get stuck into the half-hour pomodoro sesh. (Well, after ordering more coffee first. This is important.) When the half-hour is up, we go around the table again and describe whether we had any success. No one cares if someone doesn't accomplish a lot: better luck next time!

Where it goes from here remains to be seen. Ideas on new things to try would always be welcome. Maybe it wouldn't hurt to precede the pomodoro sessions with micro-workshops, short discussions on things such as good writing style, document-writing languages such as LaTeX or Markdown, why text-editors are better than word processors, and how best to use git to version control a document written in natural language. But such things would depend on the interests and goals of those attending, and whether it would get in the way of drinking coffee while defending our hands from possessed laptops.

Details:

Writing Gym is held at 8am on Tuesdays, on the alternate week to the Parkville Bioinformatics Journal Club. To stay up to date on the time and location, please join the mailing list.

Tuesday, 28 January 2014

Leading a Journal Club

Leading a Journal Club

Why lead a journal club meeting?
●                  It's a great way of having everyone get to know you.
●                  It's fun (speaking from personal experience).
●                  It gives you an opportunity to read something that you wouldn't otherwise devote the time to. Maybe something outside your field.
●                  It can give you the opportunity to get feedback and perspective about a paper that you are using in your own research. Maybe a pivotal paper in your literature review.
●                  It's a great opportunity to practise your professional skills like public speaking, summarising content and critical analysis. And you can do this all in a supportive and casual setting.
●                  That warm fuzzy feeling that you get from doing something nice that you don't have to do!
Step 1: Choose your meeting
There's a link to the schedule from this page: http://parkville-bioinformatics-journal-club.blogspot.com.au/ (top right corner)
Step 2: Choose your paper(s)
You would usually choose one or two papers. If one of the papers is quite domain specific you may like to choose a review paper as well to introduce people to the topic.
Step 3: Advertise your meeting
About a week before the meeting send the paper reference(s) to the group along the with the PDF and/or any relevant links. It also might be nice to explain briefly why you chose the paper(s). You can do this yourself via the mailing list, or send the details to the organisers and they will sort it out.
Step 4: Print the paper(s)
Come with a few print outs of the paper if possible. Usually about half the people print their own, so 4 or 5 copies is usually plenty.
Step 5: Summarise the paper(s)
Give a brief verbal summary of what the paper was about. Aim for about 5 mins. Think of this as a verbal abstract, but with perhaps even less detail. The idea is to give people who haven't read it a basic idea of what the paper was about.
You may want to finish with a comment about what you thought of the paper. What were it's strengths and weaknesses? Why did you choose it?
Step 5: Note some conversation topics
Note down a few questions or comments that came up for you when you were reading the paper. This way if the conversation lags you will have some things to bring up.
On the day:
These times are a guide only to give you and idea of what to expect. If you are leading the group, you are in charge, so feel free to work at your own pace.
7:55am Leader attempts to get there first and claim the big table at Dr Dax. 8:00am People start arriving and order coffee/breakfast. Keep an eye out for new members. They will be lurking, looking uncertainly at groups of people.
8:10am Consider doing a round of introductions, particularly if there are new members. Hand out printed papers as needed.
8:15am Leader gives a quick summary of the paper and then opens the discussion up to the group. If the conversation lags the leader should bring up one of their questions/points about the paper.
8:50am Round up the discussion. Check when the next meeting is and who is speaking (or ask for a volunteer).
Optional: Write a short blog post
If you're really keen, and you want to get even more professional development out the experience, you might consider writing a blog post. You can write about whatever you like and/or reuse your notes from the meeting.

Most of all, remember to enjoy yourself. This is not a formal presentation. No one is judging you. We are just a group of newbies trying to make sense of the field.

Wednesday, 16 October 2013

Topic posts

From time to time,  journal club participants will blog about topics discussed in a particular week. This is a purely optional part of leading the journal club, and the format and content is entirely up to the journal club member.

They may appear in an order different that of the journal club; they may also precede the week they are discussed.

The first is from Harriet: http://parkville-bioinformatics-journal-club.blogspot.com/2013/09/metagenomics-hypothesis-driven-fishing.html