Quantum Forest

As some people know, I have been using asreml for genetic analyses for quite a few years and even keep the ASReml Cookbook^§. I was quite happy to see the development of asreml-R, a package that makes available most of ASReml’s functionality from R. This made my life easier: I still use plain-vanilla ASReml for very big jobs, but I can access a much more comprehensive statistical system for fairly substantial jobs.

One of my main problems with asreml-R is that is not available for OSX (mac). Yes, I can dualboot or use a virtual box, but both options are a bit of a pain. I rather use my computer with its primary operating system and no strange overheads. I have requested several times to have a mac version. It seems that the code can be compiled without problems, but it is the license management software that is not available for the mac.

I then started looking for options to run genetic analyses. nlme was designed around hierarchical models and fitting experimental designs did not feel right. lme4 is looking good and the main issue was around fitting pedigrees, a matter at least partially solved by the pedigreemm package. I then came across the MCMCglmm^§ package, which has some nice features: it makes Bayesian analyses accessible, ready support for pedigrees and a syntax not that different from asreml-R.

After playing with the MCMCglmm library, I found that I could not use pedigrees with parents acting both as males and females. I modified the code (line 26 of inverseA.R) to print a warning rather than to stop and the compiled the library again. Voila! it is working (the beauty of having access to the source).


R CMD INSTALL /Users/lap44/Downloads/MCMCglmm --library=/Users/lap44/Library/R/2.9/library


By the way, ASReml is still my primary tool at the moment, but I enjoy having good alternatives.

For quantitative geneticists and breeders out there: ASReml runs OK in any mac with Intel processor and Parallels. You can find some comments on using it in the ASReml Cookbook.

Incidentally, SAS works OK in a mac using Parallels too. However, I am enjoying R a lot more.

It has been a log time without writing about research. New country, new city, new job. In addition, consulting and professional service. Last but not least, family and friends come first: the end result is very little time to blog and even less for writing about research.

So, what am I doing at the moment? Simple, trying to figure out areas where I am not hitting diminishing returns too quickly. For example, estimating two hundred variance components is too rich, if we can do the job with ten. The practical return from all the additional works tends to zero: we are not making much of a difference. So, what’s the point? Yes, I can publish that, but who cares?

From a practical point of view, the real issue for me is on what is affecting competitiveness in a big way. Forestry is a long term endeavour, and the longer the rotation the higher the risk. From that point of view, extending rotation because radiata pine wood quality is not good enough borders on the stupid. Doh, of course is crappy wood; answers:

1. Use something else or is there life beyond radiata pine?
2. Select and breed for trees that have decent (I do not mean good) quality.

So, what are my current obsessions?

• Profitable shorter rotations. What are the limiting factors (hint: crappy wood quality, small size pieces and scale of the operations) to make this happen?.
• Very early selection of adequate trees. Notice emphasis: selection does not to be perfect to be useful. Adequate selections at age two is much better than good selections at age ten years.
• Why do trees grow the way they do in wood properties? Why do trees choose different strategies that have such dramatic differences in wood quality?
• Rapid turn-over breeding strategies. Are we still taking fifteen years for a breeding cycle? It is 2006! Can’t we do any better?

There is an obvious quantitative void in my obsessions, I know. But I am going back to attempting to understand some basic processes before I embark in more number cruncing. Despite of this, I am also interested (but not obsessed) in the following problems:

• Simulation of breeding strategies. I have a project working on this topic starting in October this year.
• Mate allocation and population structure. Trying to show that we can get rid of sublines and other artificial groupings when using sensible mating policies.
• Large scale genetic evaluation: how simple is simple enough? My way to help having frequent genetic evaluations.

What else? I am involved in a couple of three projects with students, dealing with wood quality, breeding or both. I have a new Ph.D. student starting in August on the interaction of economics and breeding. Ah, I almost forgot: there is a large number of lectures coming my way, better look busy…

Last week I attended to the 13^th Australasian Plant Breeding Conference. This conference was somewhat tangential to my interests—which are much closer to tree and animal breeding—but it still had enough appeal to give it a go.

As in many conferences, most of the presentations where appallingly boring and/or badly done. What is with scientists and presentations that manage to turn an interesting topic in insufferable mumbling? Anyway, there were also a few good presentations and I was thinking what did they have in common? It was not the slideware, or the choice of fonts or even the number of slides. The most important part, I think, it is to have a good story. The story provides enough background to be of interest to anybody, the story is coherent and one can see how the pieces fit with each other.

For example, I have no idea about rice breeding or tomato breeding. Until last week I did not use to care about them; not at all. However, when you start a presentation saying that 3 billion people (yes, half of this planet’s population) get most of their calories from rice you really get my attention. So maybe Swapan Datta’s presentation on golden rice (see the wikipedia article too) was not flawless, but I could feel the sense of urgency and importance of this work. Susan McCouch’s presentation on understanding and using the variability of rice populations was very entertaining. Steve Tanksley had a great story on breeding tomatoes using molecular genetics tools, going back to wild relatives and skipping phenotyping for a few generations¹. So there are some scientists that are good speakers and can communicate a good story.

Note to organisers: having a lectern with a fixed microphone and the screen behind the presenter does not help to give good presentations. Some people like to move around, some people want to point things in slides so they turn, but when they speak the microphone does not pick up any sound.

¹ This reminded me of a presentation on tomato paste ideotype breeding with molecular markers, by Jeanne Romero-Severson in the Genetics of Radiata pine conference (Rotorua, 1997).

I have slightly changed the focus of my attention during the last month or so. My current obsessions are:

• Optimisation of breeding programs, for which I am learning to use AMPL, with Fritz’s help. I will need to create and format some data to include in some simulations analysed by AMPL and I think I will use Python to prototype them. If I run in to speed bottlenecks I will reimplement numerically intensive processes in either C++ or Fortran 95.
• Genetics of wood properties. After some early forages on wood properties—which finished when Carolyn changed jobs—I am back at it. John has been very welcoming and we are trying to put a couple of projects together. We should have some early results by mid next year.

There are a few bits and pieces that do not fall in these two broad areas, but they will converge pretty soon.

Family-wise

Working with Marcela and Orlando in the veggie patch. I have never had much of a green thumb, but I am really trying. We sowed coriander, parsley and chervil, and planted bok choi, onions, dill, lemon balm and capsicum. Apart from the capsicum seedlings that are struggling (a drainage problem is my guess) everything is doing fine.

Marcela’s worm farm is the old-new addition. We used to have a worm farm in Australia, but due to quarantine issues, we decided to leave it there. So we needed to get a new one plus order the first batch of worms by mail.

This has been a long hiatus without posting in the blog. My excuse? I was evaluating a number of CMS to provide members of the IUFRO unit 2.04.02 (Breeding theory and progeny testing) with a way of communicating, including web page, newsletter, forum, calendar of events, etc. As the coordinator of the group, I am really keen on having people interacting, but the IUFRO site is, to put it mildly, close to useless.

I tested Drupal, E107, Mambo / Joomla and Plone. The first four are written in PHP and require PHP+MySQL, while the latter is written in Python. Although I prefer Python as a language, the idea of getting special hosting for Plone, because of especific server requirements is a bit of a turn off. I first had a quick drive test of lots of CMS at the Open Source CMS site, which is a great resource. After that I installed and configured Drupal, E107 and Mambo in my test system, where they are still working together without any conflicts. CMS Matrix provides a fairly detailed comparison between all the systems.

In principle the simplest interface seemed to be Mambo’s (plenty of eye candy), followed by E107. Initially I struggled with Drupal’s interface, but after a while I got the hang of it. On terms of functionality, all systems seem to provide most of the features I need: basic user management, forum, event coordination and easy posting. However, after a while I found that Mambo/Joomla are a bit of an overkill, bringing too many things by default. Putting together a magazine-like interface in E107 implied creating folders by hand (one per issue) and posting in those folders (a bit too primitive for my taste). Going back to Drupal, I started exploring all the contributed modules and I think I can get everything that I need from there.

I am now slowly putting together the IUFRO Breeding theory and progeny testing site, which should be operational by January 2006. The configuration of Drupal will be explained in my next post.

Part of my two weeks away included a meeting with old friends in Noosa, Queensland. Noosa is a very civilised place where to discuss breeding strategies: warm even in winter time, with a lay back, holiday atmosphere. After each day of discussions there were plenty of chances for having long walks followed by dinner. We stayed at the Noosa Lakes Resort for the third time (location map).

The picture shows Mark Dieters, Colin Matheson, Heidi Dungey, myself, Tim White, Jeremy Browner, Fred Burger and Mike Carson. In addition to people in the picture, we had Paul Jefferson, Michael Henson and Steve Verryn (next to me in the picture below) in the meeting. Colin—our resident wine buff—made some interesting choices so we tried a wide range of whites and reds.

It was great to see Tim again, after all these years. I met him for the first time in 1993, when I thought ‘this is a very clever guy’: pretty good at navigating the politics of meetings. He did not disappoint me and this time he was even better, helping us to come up with a good strategy.

I am looking forward to participate in other strategy meetings. With some luck Steve may be able to organise a conference in South Africa (country that I have never visited) and we could have a go there.

This post represents the first official reference to Plus Tree, the site that will support most of my genetics and breeding work. I know, the site is not yet ready for prime time, but I think it has reached a state of ‘potential usefulness’ for people interested in the topic.

I will try to post short essays around once a month, but there are no guarantees. A friend of mine asked ‘Why are you giving away that stuff for free?’ The simple answer is because most posts do not represent ground breaking ideas; however, the fact is that they are rarely implemented in breeding programs — particularly in a single program. In addition, imagine what you would get if you actually pay me to do some work!

In the not so distant future I will make available a couple of features that I am still testing for the site, including access to open projects — or at least the basic ideas behind them — as well as (probably) a Wiki system.

Some people know that I have been looking from the sidelines of genetics for a while. Let’s say that now I am starting to play again: slowly but steadily.

I finally completed (and submitted to Silvae Genetica) the manuscript for ‘Genetic variation of physical and chemical wood properties of Eucalyptus globulus‘. This is not my first or last paper (it is publication 25), but it took such a long time that it deserves a special mention. The project was plagued with problems and delays that, although did not affect the final quality of the data, made data analysis and writing the manuscript a real pain in the back.

By the way, Silvae Genetica looks like a very old fashioned 1800s journal. I always associate the image of a very old German worker printing the journal in a damp basement. Nevertheless, it is almost compulsory reading for tree breeders and the publisher seems now keen to give it a facelift.

When writing papers I use either a combination of MS Word and Endnote (a reference manager) or LaTeX in its MiKTeX incarnation with TexnicCenter as a text editor. I use the latter combination for large documents, like convoluted course notes. This time I chose Word but did not have a ‘Silvae Genetica style’ for Endnote, which is necessary to format the citations in the text. I created a style that works for journal articles, books, book chapters and conference proceedings, which you can download from here.

I do not expect to see the manuscript for around three months. By then I should receive comments (I hope positive) from the referees.

It has been fifteen months since I stopped working full time in genetics. I was feeling tired — some may even say burned out — of the topic. After ten years there was nothing really exciting. I think that I was trapped by the routine, dealing with small problems and loosing track of the big picture. I did not completely stopped working on genetics, but I have been spending less than ten percent of my time on it.

Two weeks ago I started thinking again on tree breeding, but mostly about big picture items. For example, how to integrate different parts of breeding programs rather than how to obtain an infinitesimally small covariance component. I realised that I still enjoy very much this area of genetics and breeding.

This experience got me thinking about the rise and demise of research groups. During the 1970s the tree breeding group in the New Zealand Forest Research Institute (now Forest Research) was one of the most exciting places to be in, putting forward new breeding strategies, using selection index and type B correlations. Later, during the late 1980s and early 1990s, the School of Forest Resources and Conservation from the University of Florida was a very interesting place, and they were busy applying BLP (basically selection index) to every possible dataset. Starting in the mid 1990s, the School of Plant Science and the Forestry Cooperative Research Centres based at the University of Tasmania thrived evangelising forest geneticists about BLUP and breeding objectives. All those research groups are still active, but in general they are not at the forefront of tree breeding anymore. Some researchers will apply the same techniques to 300 datasets in one go rather than to a single one, but that is not exciting or very rewarding either from a research or an industrial viewpoint… It may be that researchers got tired after a wild ride.

Another interesting point is that many of these groups developed a reputation applying techniques developed in the animal breeding world (e.g., selection index-BLP, BLUP, breeding objectives). It is not that people in forestry are not very creative, but we do lack critical mass when compared to animal breeders. It may be that the new exciting topics will come again from animal breeding. It may also come from other areas or even be developed from the inside. I am just happy to have recovered the “love for the trade” again, and am looking forward to dedicate part of my time to work in “big picture” items, wherever they come from.