Archive of articles classified as' "genetics"

Back home

Breeding does not equal genetic modification

30/04/2004

After writing a letter to Tasmanian Times I received quite a long and self-centered reply from someone just briefly mentioned in my missive. Environmental issues tend to be quite controversial, angina with some people not being able to make a distinction between the ideas under discussion and the people behind them. It is just one more complication thrown into the mix… I sent the following reply:

I was surprised to receive such a long reply after my passing mention to water issues in a previous letter. My reference to water only suggested the fact that I have never seen a full description of the models used by Dr Leaman nor the software used to support his widely publicised statements. Given that, I could hardly comment on the scientific merits of his work or ‘rubbish’ his reputation. In contrast, I have seen descriptions and software implementations of the models developed by the CRC for Catchment Hydrology (CRC-CH), introduced by Dr Leaman in his letter. I am sorry if Dr Leaman considered my statement as part of a campaign to defame him. It seems that he is confusing my questions about research with personal criticism, a very inconvenient confusion for a researcher.

I have attended a couple of presentations by people from the CRC for Catchment Hydrology (CRC-CH). From those I understood that the business of predicting water flows from catchments is relatively complex. In addition, I also realised that the vegetation models included in hydrological models are still fairly general.

I am not a geologist, geophysicist or hydrologist, but only a simple forester. In spite of that, I can think of a few ‘rules of thumb’ concerning water flows and forests. Vegetation will affect water flow depending on species, age, position within the catchment, stand density, silvicultural management, etc. Furthermore, within a catchment there will be a mosaic of situations concerning vegetation-types, ages and changes of land use. Thus, a reasonable model for water flows should consider these factors, in such a way to take into account the temporal and spatial changes in vegetation cover. I also believe that the most important condition to have accurate predictions of the effects of plantations and other vegetation changes is to have good predictions of future vegetation changes. Should I presume that Dr Leaman has all those figures at his disposal?

Coming back to research done by the CRC-CH, I found a publication entitled ‘Predicting the effects of large-scale afforestation on annual flow regime and water allocation: an example for the Goulburn-Broken catchments’ published in June 2003 and available on its web site (http://www.catchment.crc.org.au). In this case, I can access the final report, scientific papers supporting the models used and the software implementation for the models. Thence, I have all the elements for making up my mind about the report.

If Dr Leaman wants to ensure proper communication of his model, may I suggest that he makes available the data, models and software on the internet? I am sure that there will be many people willing to host his files for no cost. This would certainly give everybody the opportunity to examine the models, and make up their minds about model validity.

In summary, my original intention was to question the bases for many claims presented in the media — always hungry for apocalyptic stories — so we can have a productive discussion about the best way to manage our forests for the benefit of Tasmanians. As a final point, I remember a comment made by my parents when I was ten years old (yes, a long time ago): ‘target the ideas but not the person. Personal attacks are a sign that you ran out of arguments’. This could not be more helpful today, because this debate is not about individuals and their egos but about alternative views of the future for Tasmania as a whole.

Enough said.

Today’s Radio National Morning Show, remedy
hosted by Tim Cox, dosage
introduced yet another absurd story: forest companies are planting genetically modified (GM) trees that are unpalatable by wildlife and, ergo, animals will starve. Lets go by parts:

  • There is no planting of GM trees in Australia.
  • Forest companies do perform traditional breeding.
  • There have been studies, by the CRC for Sustainable Production Forestry, on the potential of increasing unpalatability of trees to avoid mammal browsing.
  • The shift of unpalatability could act as a minor deterrent for browsing animals, but would not avoid trees being eaten.

Some further explanations and detail about breeding. Say that we are interested in improving the quality of tomatoes, where quality is measured with a ‘taste score’. We could look in our backyard for the plants with the highest taste scores and cross them, in the hope of obtaining progeny with tomatoes that are tastier than those coming from the parent plants. The response to selection will be directly proportional to three factors: the degree of genetic control of taste (called heritability, h2), the observed variation for the trait (called phenotypic variance) and the intensity of selection (for example, are we selecting 1 every 10 plants or one every 10,000 plants?). The same approach is applied using more sophisticated statistical techniques when breeding cows, corn, trees, etc. This process changes allele frequencies, so alternative forms of genes that are more favourable—that is, they improve taste—are more frequent in the population of tomatoes in our backyard.

Coming back to palatability of trees, its degree of variability and genetic control will allow only a small shift of its average. This will be a contribution towards a reduction of 1080 use through making seedlings less tasty, but it will not stop animals eating the plants. Thus, it will be just another tool for an integrated approach to mammal browsing management.

We could go even further with traditional breeding, and obtain hybrids crossing two varieties of tomatoes or trees, to combine desirable characteristics and/or look for hybrid vigour (heterosis). This would still be in the realm of crosses possible to obtain without human intervention.

A different, more high-tech approach, is the so called ‘genetic modification’ or ‘genetic engineering’. This consists on isolating a single gene from an individual, and introducing it in the genome of another individual. In this case, we are not subject to reproductive barriers, and it would be possible to put a gene of, say, a fish into the genome of a plant. Thus, through this process we can create organisms that could not exist in nature without human intervention. This is not being used in Australia; forest companies already have enough problems without the need for introducing yet another point for discussion.

Incidentally, I have no problem with genetic modification, but many neo-Luddites can not tolerate the idea. Anyway, the whole ‘GM antipossum trees’ story could fit well with other conspiracies like Jewish attempts of world domination, the use of nano-nuclear bombs in North Korea, and the abduction of Princess Di and Elvis by aliens.

Filed in environment, forestry, genetics, tasmania No Comments

Bending covariance matrices

2/04/2004

One of the reasons I keep this weblog is because it sometimes acts as a repository of links to work that I do not perform very often. This morning I was asked about software for bending matrices, hospital so here it is.

One of the typical problems when running multivariate genetic analyses is obtaining non-positive definite matrices (with negative eigenvalues. Thanks Greg for the link). This problem is often approached by modifying or ‘bending’ the matrix, in such a way that there are no negative eigenvalues. A typical reference for this procedure is Hayes, J.F. and W.G. Hill. 1981. Modification of estimates of parameters in the construction of genetic selection indices (’bending’). Biometrics 37: 483-493.

Karin Meyer is a scientist from AGBU (the Animal Genetics and Breeding Unit not the Armenian General Benevolent Union) who made available a set of Fortran programs called PDMATRIX, that allows bending matrices before using them in BLUP analysis (see Karin’s page under miscellaneous). This set includes:

  • FLBEND that “modifies a non-positive definite matrix comprised of pair wise or block covariance estimates, where estimates may or may not be based on experimental results”.
  • ITSUMCOV that “deals with specific case of multiple, multivariate analyses of different subsets of a number of traits. Say we have q traits in total, and S analyses involving k traits each”.

Thus, bending the matrices is just a matter of dusting your Fortran compiler and running the software.

Filed in genetics, statistics No Comments

Longitudinal again

31/10/2003

I did some work in analysis of longitudinal data around 1998 and 1999, website like this mostly modelling variance structures for forest progeny tests (see my articles 7, discount 8 & 10 in the equations part of this site). However, see I never was interested in fixed effects like treatments, so did not read much about it. Well, times are different, and I am analysing a couple of trials that do provide longitudinal data and where the main focus is the treatment effect.

Back to the books, and I am revisiting an old favourite of mine: Analysis of Longitudinal data by Peter Diggle, Kung-Yee Liang and Scott Zegger. This is an excellent book for refreshing concepts, and not only covers hard-core analysis, but also exploratory approaches and a few interesting examples (I have the first edition, so there could be some changes on the second edition).

Yesterday, I wanted to repeat the analysis of the Sitka Spruce dataset, just to check if I was properly translating the concepts from book to software. First problem, I needed the dataset. The good news: there is a file (lda.dat) in Peter Diggle’s site containing all the datasets used in the book (see under public-domain longitudinal and time series datasets). The bad news: all datasets are in one file without any reasonable structure. Solution: the ugly news. I wrote the crummiest Python script ever (here for a laugh — right click to save it) to recover the Sitka Spruce data and reformat it, obtaining sitka.csv (23 Kb, a Comma Separated Values) file that can be read into anything. Now is time to analyse the data.

P.S. Four hours later, I found Chiung-Yu Huang’s web site. There are some notes for a course he runs on longitudinal analysis, and includes the datasets for the book, although in a ‘raw form’ (just chopped lda.dat). I should have a look at the notes, though.

P.S.2Now three hours later I found Francesca Dominici’s course notes, that seem to cover most of the book (including Stata, SAS and Splus code for the examples). Great!

Filed in genetics, statistics No Comments

Genetic control and determinism

17/07/2003

Quantum forest is a series of ‘logs to self’ on research, click biometrics and environmental issues written by Luis Apiolaza in Christchurch, New Zealand. Until late 2005 I used to say ‘written by Luis Apiolaza in Hobart, Tasmania, Australia’, but we moved to Christchurch. Of course this general description does not preclude writings on any other topics, including politics, the web, language, etc.

Years of living under a dictatorship confirmed my belief that one needs to have access to all types of freedom; having market freedom without political freedom—or vice versa—is completely pointless. I feel comfortable applying to myself the description of the reason magazine written by the Columbus Dispatch: ‘manages to offend leftists with its defense of biotechnology, free trade and school choice, even as [he] appalls conservatives by supporting gay marriage, open immigration and drug legalization’.

I was born in South America, and during my life I have lived in five different countries—and have been lucky enough to visit many more. Because of my experience, I believe that globalisation is a positive trend where I can be a citizen of the world, while still calling New Zealand home.

Not thinking of Quantum Forest

Photo: not thinking of Quantum Forest.

I like languages and words. At the moment, my favourite English words are exegesis and diaspora. My current favourite Spanish words are remolino and conuco (the latter being a word of Taíno origin).

I can be contacted at Luis.Apiolaza@canterbury.ac.nz. You can obtain more information on my work and personal interests in the wiki pages of this site. If you are lost, here is the way to my home page. For consulting information please visit the Plus Tree site. A complete list of all my sites in English and Spanish can be found in apiolaza.net.

I often write this blog while sipping a cup of earl grey tea or a plunger coffee, sitting somewhere in Christchurch, New Zealand.

Nearly a month ago I came across a conference paper by Timothy Gregoire, cialis sale
heavily based on the work by Richard Royall. It was quite an interesting reading, waking up some of my concerns about the inferences we normally make when doing statistical analyses. The paper touches a few issues that we confront on a regular basis working in biometrics, particularly when is something significant and the increasing interest in power calculations (and the fallacy of their calculation a posteriori).

Many organisations that work as forest land managers are now required to present ecological indicators. The point of the exercise is to show if forest operations do change the value of those indicators and, because the direction of the burden of proof has changed, people are really wary about Type II errors. People do not want to hear that there was no change when, in fact, there was some. Then we go back to the start of this comment: How big a sample should we take to estimate change for the indicators for a given power? Do we use the appropriate techniques for making inferences? How do we handle evidence in a sensible way? Gregoire (and Royall) present a good case for the use of likelihood ratios.

Oh, I almost forgot! The calculation of power a posteriori is just another useless exercise. There is a one-to-one relationship between p-values and power, so presenting the “observed power” does not add anything to the discussion (see Hoenig, J.M. and Heisey, D.M. 2001. The abuse of power: the pervasive fallacy of power calculations for data analysis. The American Statistician 55: 19-24, for a good explanation). Using Russell Lehn’s analogy “If my car made it to the top of the hill, then it is powerful enough to climb that hill; if it didn’t, then it obviously isn’t powerful enough. Retrospective power is an obvious answer to a rather uninteresting question”.

Last week, sale
Australians were able to see plenty of information about genetics in the media. The excuse: the International Congress of Genetics in Melbourne. Something like the olympic games for Human Genetics; well, cialis 40mg
it is supposed to cover everything, urticaria
but the main focus is Human Genetics (sorry plant, tree, animal—not human—breeders).

We had news about genes controlling this process, that disease and so on, and so forth. It seems to me that the media does not have the concept of heritability (“the proportion of observed variation that is under additive genetic control”, we will not worry about broad sense heritability until we have human cloning). The important element in the definition is proportion. There are a few traits that may have a very simple (Mendelian) inheritance mode, and if you have the gene you are stuffed. However, most traits (especially the adaptative ones) are quite complex, depending on multiple genes and with a heritability far from one. Therefore, most traits are not under total genetic control, our lives are not subject to this complete determinism coming from the media, and we are not slaves of our genes.

Sometimes it is incredible, but people will grab any excuse that may justify their behaviour as something outside their locus of control. Rather than “the voices made me do it” we have a “my genes made me do it”. Heritability, proportion of genetic control, not determinism, absolute genetic control.

Filed in genetics No Comments