Thursday, April 16, 2009

Cladistics Workshop Announced


The Ohio State University and the Willi Hennig Society have just announced this summer's Workshop in Phylogenetics Indoctrination in Cladistics Workshop. Some twenty students will receive fellowships to attend this workshop from the Willi Hennig Society. With these fellowships, students will be able to receive four days of instruction on the proper use of outdated methodologies for only $600. On the fifth day, the workshop will make a foray into the 21st century with a lecture on model based phylogenetics and a laboratory on the use of RAxML, Garli and MrBayes. Instruction on model-based methods will be provided by Dr. Christopher Randle, whose only publications on Bayesian methods are critiques (1, 2) and whose recent publications rely either exclusively on parsimony (3) or give preference to parsimony over maximum likelihood when the two methods are largely congruent (4). I'm sure Dr. Randle is an excellent scientist, but his presence as the sole instructor of model-based methods suggests that this workshop is going to be about as balanced as Fox News.

28 comments:

  1. Will it include Farris in overalls? If so, I'm going.

    ReplyDelete
  2. I'm not sure it's meant to be balanced...might have made more sense to leave out the ML and Bayes altogether here and let students learn that at Woods Hole or Bodega Bay or...class? Will surely be the most in-depth foray into cladistics most students would ever get - anywhere!

    Have to say that the pic was funny, though!

    ReplyDelete
  3. The first paper you mentioned is a dead link. However, I've linked to them on my site.

    But you've missed another paper.

    In 2005 Pickett and Randle wrote a critique of Bayesian methods claiming that the inherent unequal priors on clade size, due to using a uniform tree prior, biased posterior probabilities (1), after which Dan Warren, Adam Leache, Jim McGuire, and myself wrote a critique demonstrating why they were wrong (2). The 2006 paper you link (the dead link) is a rebuttal to our critique(3). We have another paper in press at Syst Biol (proof with corrections, 4).) where we revisit some of this "debate" (although the scope of the paper is not an outright critique of Pickett and Randle).

    I was unaware of the other 2005 paper.

    While I'm on the subject, Pickett and Ward Wheeler wrote yet another paper advertising their misunderstanding of Bayesian methods (5). which was also severely beat down by two of Rafe Brown's grad students (6).

    ReplyDelete
  4. Also, if you want to see a stealth photo taken of Farris in his overalls, you can get that here: http://pantheon.yale.edu/~mcb64/pubs/farris1.jpg

    I took this at the "constructing large phylogenies" workshop at Ohio St in 2005. The entire thing was more-or-less an unadvertised Willi Hennig Society meeting with some token statistical phylogeneticists invited so they could be criticized.

    Some highlights:

    -Kluge getting all red-faced and angry.

    -Farris would sitting in the back of the auditorium basically take up two seats. When a speaker said something he disagreed with, he would laugh out loud *exactly* like Jabba the Hutt.

    -A borderline unethical episode where a grad student gave a talk on using TNT to construct large phylogenies. This was immediately followed by a seminar by Goloboff with was 30 minutes of "why the guy who just talked before me is an idiot". The thing is, it was not set up as a debate (i.e., the first guy had no idea he was going to be critiqued and thus could not respond).

    -Also, I seem to recall Ward Wheeler making the "how can we really *know* anything?" argument. In other words, the same philosophical discussion you used to have while smoking a bong in your freshman dormroom.

    ReplyDelete
  5. Awww, it seems that more than one person here has psychological scars after interacting with members of the Hennig Society.

    But don't despair! The society will soon create a fund for the treatment of victims mentally abused by their own members. ;)

    ReplyDelete
  6. The horror! I saw that picture in a live version on the society's meeting in Brazil.

    Also saw Farris almost getting to fisticuffs with Swofford at Smithsonian.

    Good stuff. I guess in a couple of decades they will start using ClustalW, and then nodody will stop them.

    ReplyDelete
  7. ... it is better to say "is going to be about as balanced as this blog." xP

    ReplyDelete
  8. Unbalanced: Refusing to use and endlessly criticizing demonstrably superior methodological approaches for quasi-philosophical reasons.
    Balanced: Being open to the use of a complete range of methodological approaches, and selecting the one(s) that seems most appropriate for the specific question(s) at hand.
    Humorous: When balanced scientists make fun of unbalanced scientists.

    ReplyDelete
  9. Because comparing anything you dismiss to Fox News = humour?

    ReplyDelete
  10. Gwynne, try typing "Fox News" into that little search box at the top of the blog, then hit enter. I think you'll find little evidence to support your hypothesis that we compare "everything we dismiss" to Fox News.

    ReplyDelete
  11. Hi,

    Is this the phylogenetics discussion forum for seven-year old kids? If so, I'll like to register my son.

    Thanks.

    ReplyDelete
  12. This is a post from Pablo Goloboff (I posted as "anonymous" only because I don't have a google account).

    I apologize for the length of this post, but there's a number of facts to be corrected, regarding Matt Brandley's account of the "borderline unethical episode" supposed to have taken place at the MBI meeting in Ohio, 2005, with me, Pablo Goloboff, as the main culprit.

    There are a few details Brandley left conveniently out.

    The "grad student" --Usman Roshan-- already had his Ph.D. and a Job at the New Jersey Institute of Technology. He was not alone; his coauthor Bernard Moret was there, too. Moret was at the time the head of the CIPRES, funded by NSF with over 11 million dollars. This was not just a poor loser or a pair of idiots. If there was a loser in the room, it was me, working in Latin America, with no NSF grants, and a salary of about 600 US dollars a month.

    According to Brandley, this poor "grad student" gave a talk on "using TNT to construct large phylogenies". Yes, Roshan, Moret and collaborators (Roshan et al., 2004. Rec-i-dcm3, a fast algorithmic technique...) had used TNT to analyze data sets, but they did this only to show that their own method, rec-i-dcm3 was superior:

    "Our study shows that
    Rec-I-DCM3 convincingly
    outperforms TNT —the best
    implemented MP heuristic—
    often by orders of magnitude,
    on all datasets"
    (Roshan et al., 2004)

    In case you haven't noticed (yes, I know I have to explain things carefully because some of you are likelihoodists): that rec-i-dcm3 is *superior* necessarily implies that TNT is *inferior*. But the superiority of rec-i-dcm3 was a result of unfair comparisons and the misuse of TNT. Which is what I showed in my talk.

    Both Moret and Roshan later acknowledged that their behaviour in this matter had not been entirely appropriate. They did this during mid 2005 --well before the Ohio meeting Brandley fantasizes about. Which brings me to Brandley's last lie:

    "the first guy had no idea he
    was going to be critiqued and
    thus could not respond"

    the guy KNEW PERFECTLY WELL he was going to be critiqued, because we (him, Moret, and myself) had exchanged many emails about it, *before* the Ohio MBI meeting. And the next email of Roshan to me started by saying:

    "Hi Pablo,
    How are you? Hope all is
    well. It was finally good
    to meet at MBI. I actually
    did some more runs with
    recidcm3 using ..." (etc)

    which is to say: Roshan himself did not feel he had been treated like an idiot.

    That's for the facts. Now, only two short non-factual paragraphs...

    The first: I just wonder how you manage to cope with the ethical dilemma: treating Roshan like an idiot would have been wrong, but mocking cladists and having fun at their supposed idiocy is a lot of fun. Hmm, maybe, just maybe, you are not so nice and honest and balanced after all?

    In any event, who cares about that; what matters is how you feel, right? And I am sure you feel G-O-O-D, and you will all continue having fun caricaturing cladists, and feeling oh-so-smart and cute and modern and fashionable and, of course, balanced, because the "methodological approaches" you use are "demonstrably superior".

    OK, enough of my time wasted in this blog already.

    ReplyDelete
  13. Many people equate cladistics with the parsimony method; I guess as a "North American Workshop in Cladistic Methods", you should expect to get a lot on the parsimony method. I imagine that students will receive a first-rate introduction to the latest advances in parsimony, which seem to focus mostly on improvements to tree searches.

    I tend to look at cladistics differently, however, and think that the workshop students will suffer from not learning about explicitly model-based methods. It is becoming more important for evolutionary biologists to understand these methods. Graduate students who don't understand continuous-time Markov models, maximum likelihood, and Bayesian analysis risk missing out on what I think are more appropriate methods for data analysis, or worse yet, missing out on job opportunities. (My experience is that universities hire people who can sensibly teach model-based methods.)

    ReplyDelete
  14. 1. I didn't realize that a critical examination of Bayesian analysis made one a cladist. Joe Felsenstein will be disappointed to find this out.

    2. Regarding teaching methodology, I'm a total pluralist. Students had better be well versed in parsimony, ML, and Bayes if they want grants and jobs.

    3. In the situation Matt recounts, being set up for public embarassment is "borderline unethical". If this happened (which I doubt it did), I'd certainly label it "juvenile". I would apply the same term to an anonymous blogger for attacking a scientist on some home-cooked web forum over a perceived difference in professional opinion.

    4. Model based methods are winning, right? So why do you guys have your panties in such a bunch? It makes your posteriors show*.

    *OK...that's ridiculous, but come on. Grow up.

    ReplyDelete
  15. It is disappointing to read these comments because the model-based approaches already dominate our community, so parsimony methods we will discuss at our workshop cannot present any threat to those uninterested in using them. The comments show that, in addition to not teaching our methods, we are also not supposed to teach people to use model-based programs, either. What can be the point of these screeds? The contributors to this blog make clear that parsimony procedures (and their advocates) must be exterminated, and with glib dismissal. Who is biased here? For the record, although there has been discord among visitors to our workshops, the ML and the MP people among the faculty and students at Ohio State all get along well, we teach together, and we do not see that a scorched Earth policy toward competing methods is either a scholarly opinion or operationally useful. The critics who have posted comments here could stand to learn something from us.

    John W. Wenzel

    ReplyDelete
  16. This is a bit like writing about harvesting embryos from Blaxican Hasidim creationist abortions in a Chinese prison for homosexuals and political dissidents.

    It's not about who's right or wrong so much as it is about the terms being grating and outdated, along with the topic.

    ReplyDelete
  17. Wow, this has certainly brought out some interesting discussion and perhaps the not-the-best part (I hesitate to say worst since that could be pretty far down) in people. Despite the original post's dipping a bit over the line, it did seem to be trying mostly to point out that calling a workshop that is 80% parsimony with just a quick overview of model-based methods a "workshop in phylogenetic methods" is a bit of a stretch.

    Please do remember that anybody can make comments and we rarely remove them unless they're totally off-topic...but that doesn't mean that the authors of the blog are all of the opinion of all the commenters. In my opinion, this has gotten a bit out of control and nasty... and I hate to see that take away from all of the good discussions that have been going on here recently.

    It seems that there is plenty of room to have a constructive, intelligent and *mature* discussion about different methods - and hopefully this exchange will have the silver lining of broadening the readership base so that we can have that.

    Can we move on, please?

    ReplyDelete
  18. [...]There is a skirmish going on at Dechronization blog right now. This is a coauthored blog about phylogenetics[...]

    ReplyDelete
  19. Apologies to Dr. Goloboff for the appearence of unsubstatiated allegations in the comments section of this blog. I'm grateful that you took the time to respond and sorry that you had to take time out of your schedule to do so.

    The authors of this blog have diverse opinions and perspectives, but we all seem to agree that this effort at humor was taken too far. Feel free to continue commenting, particularly if you feel like your views have been misrepresented, but I'd like to us return our attention to the scientific issues that are the reason for this blog's existence.

    ReplyDelete
  20. @ Susan Perkins
    A bit over the line?

    ReplyDelete
  21. I was at OSU in 2005 as a graduate student and remember the scene as described by Goloboff.

    As far as maturity is concerned, the initial blog sets the tone by stooping low, singling Randle out from all other participants and questioning his qualifications and ability to objectively teach statistical methods. Others continue this incredibly mature string by poking fun of Farris' fashion sense, commenting on the condition of Kluge’s facial coloration, and distilling Wheeler’s sentence down to a pot episode. But as soon as one of them should comment on their own behalf, the conversation has officially turned immature?

    And what editorial cover-up of the century was exposed when it was discovered that, gasp – the Willi Hennig Society was teaching a cladistics workshop? Form your own methodological society and hold your own workshop. If ML and Bayes are the way to go then soon enough we will be eaten by our young and you will have the world at your fingertips with the last remaining cladists groveling at your feet for your phylogenetic scraps. Until then, in the spirit of academia, allow students who want to learn about cladistics learn about cladistics.

    ReplyDelete
  22. @Glor-- You were feeling congratulatory over those unsubstantiated comments a few posts back. What changed?

    ReplyDelete
  23. The fact that we've heard another side to the story.

    ReplyDelete
  24. @Glor
    Are you in the habit of assuming that all unsubstantiated claims are true?

    Did you not interpret the rest of the post as an exaggeration?

    ReplyDelete