New content for asreml-r is available here: http://apiolaza.net/asreml-r
Contents of the ASReml Cookbook:
- Obtaining ASReml and its manual
- Getting help
- Using a decent text editor
- Using ASReml in a mac
- Take me to the recipes!
- Improving the cookbook
- Citing or referencing the cookbook
The cookbook is also available as a Single Page, which makes easier to print it as a single document.
ASReml is a very powerful statistical software developed by Arthur Gilmour from New South Wales Agriculture, with support from an international team of leading statisticians. It is specifically designed for fitting mixed models for large datasets, with unmatched model flexibility and high speed. It is far faster than SAS, S-Plus and similar software. I have a few reasons why I think that ASReml is the best software for genetic evaluation.
ASReml’s manual is the source of all knowledge, with over 200 pages. That is a bit of an overstatement, but it is certainly exhaustive and sometimes daunting. One of the problems when you start using ASReml is how do I start? How do I setup my models the way I used to in SAS / Genstat / DFREML / etc? I have trees, no animals, what do I do? I have spent a long time learning how to fit models with the software, and I have I decided to make some of the information available in this site, in case it can be of some use for you.
How do I obtain ASReml?
ASReml is a commercial product. Trial (30 days) versions can be obtained from VSN’s International web site. The page Obtaining AS Reml provides full instructions for getting the software and the manual.
There are several versions or ways to access ASReml:
- It can be used as a console (terminal) program, writing all the code in a text editor and then using the command line to submit the job. This is the oldest way, and the one that I have used the most.
- It can also be used as a windows program with its own editor (at least under MS Windows). I have never used it this way, which I consider mostly a hassle.
- It can be used as a library or module called from R or S+ (asreml-R). This has the convenience of using a general purpose statistics software for data preparation, plotting and further processing, combined with the power of ASReml. In this case ASReml acts as an alternative to the lme library, with efficient processing of data sets with a pedigree. This is the way I am moving towards at the moment. I will add more information on using ASReml from R during 2009.
When buying a license the user gets access to all three versions.
The core FORTRAN functions of ASReml are also the engine of the REML procedure in GenStat. However, this site does not provide any information on using that software.
Besides the manual, the cookbook and other web sites, there is now a forum in the VSN site. Please do not contact me for help with your analyses, as I do no have time for that.
Using a decent text editor
To run an ASReml job you need to write an ASReml command file (the file with a .as extension). Hence, one of the first things you should get is a decent text editor. If you work in MS Windows you already now that Notepad is hopeless, and probably you do not want to spend any money. There are plenty of free or shareware editors around, but I recommend using Context, which is a free ‘ASREml aware’ programming text editor. There are a few instructions to Setup Context for ASReml use.
Can I use ASReml in my mac?
Yes. Starting in December 2009, ASReml is natively supported in OS X (both 32 and 64 Bits). However, there is not yet an official asreml-r version; there is, however, an experimental asreml-r version.
In addition, there are two other ways of running ASReml on Intel-based Apple computers (Macbook, Macbook Pro, Mac Pro and new mac minis and imacs):
- Use Bootcamp to boot your mac in one of the operating systems supported by ASReml. For example, you could boot using Windows XP and the mac would essentially be a windows machine.
- Use a virtualization machine, for example Parallels, to have multiple operating systems running on your mac. In this case there is a windows containing a session of Windows (or Linux, or whatever) running inside OS-X. This is how I run ASReml in my Macbook Pro.
I have tried this only with the Windows version of ASReml, but I assume that it would work the same with the Unix versions. I now prefer dual booting, because the system is much more reponsive (and a mac is a very good windows machine). If you go for the second option be aware that you will require plenty of RAM; essentially you will be running two operating systems at once. With 1GB my system felt sluggish but it worked quite well with 2GB and higher.
Here is what you are really looking for! Below there are recipes to run the following types of analyses, the first link is for ASReml console (or classic) and the second link is for asreml-R:
- Job Anatomy
- Basic structure
- Equation model description
- Univariate Analysis (R version)? including:
- Some basic models
- Equivalent models
- Clonal data
- Multiple site, single trait
- Covariance Structures (both versions).
- Multivariate Analysis (R version)?:
- Single site, multiple traits
- Two sites, single trait treated as multivariate
- Estimating Heritability:
- Functions of variance components and .pin files
- Genetic correlations
- This section is in preparation and is incomplete
- Longitudinal Analysis:
- Longitudinal analysis in ASReml
- Random regression models
- Spatial Analysis
- Spatial analysis in ASReml
- Spatial analysis for irregularly spaced trees
- Data Tips
- Truncating the dataset
- Univariate to multivariate format
- Starting values: the apocryphal Dutch method
- Simulations Using SAS
- Simulations Using Python
- General Resources
- Other ASReml sites
- Race classification
- General information produced by the CRC-SPF
- Cooking Recipes, including some real recipes to avoid hunger while running ASReml
- Home made pizza dough and toppings
- Beef with plum sauce and Chinese greens
- Smoked chicken and cream pasta
The cookbook also includes interesting replies to ASReml discussion group posts. Note: I have not necessarily used or checked the advice:
- How do I fit Generalised Models?
- Binomial and Poisson models
- Problems with this type of analyses
- Problems Overlapping Design Matrices in genetic models
- More coming soon.
Improving the cookbook
Feel free to contact me if you have any suggestions about material that should be included here or, even better, if you have some code examples that you would like to share. If you have any problems reading these pages see the colophon.
Citing or referencing the cookbook
If the ASReml Cookbook was useful for your analyses, you may cite it in the following way:
- Author. (Date of publication, use last modified date). Title. Retrieved date of access, from URL. For example:
- Apiolaza, L.A. (February 2006). ASReml cookbook. Retrieved 18 February 2006, from http://uncronopio.org/ASReml/.
Copyright (1997–2009) by Luis Apiolaza, some rights are reserved.