updated 26 Nov 2005
in construction!
Microarray analysis
Mention the word "microarray" to a biologist and there's a high probability they recoil with fear.

The nightmare of thousands of coloured spots, plots that look like a swarm of midges in the Scottish highlands, logarithms and (shiver)
statistics! Oh, and expensive software, just in case the cost of the arrays and reagents wasn't enough.

That was pretty much my position not long ago. I did all I could to avoid getting into microarray work. But in the end it became obvious that
microarrays can be very very useful.

Whilst I am no expert, over the past year or so I have tried and tested a number of different protocols, programs, attended workshops, and I
am starting to feel confident enough and even enjoy looking at microarray spots (somebody save me! ;-)

A number of people in my institute are becoming interested too, and I decided to compile the bits of info and protocols that I feel work
well, and a procedure to do simple analysis in a painless manner. I am still writing it. If you want you can download a copy here of the "in
progress" PDF.

    Download PDF: simple guide to cDNA microarray hybridisation and analysis.


The guide offers an outline of the procedure and shows links fo several websites I have found interesting. Those links can also be found at
the bottom at this page.
When it comes to scanning slides, I have used an Axon GenePix 4200AL as an example. There is a little section on using GenePix 6.0
software to segment and quantitate the images, just because it comes with the scanner. GenePix is the only commercial software I have
mentioned in the guide. There are very good alternative
free software around, and although sometimes the interface can be a bit "clunkier"
than their commercial counterparts, they are still easy enough to use and they are getting better and better with each new update. In
addition, as there are no license restrictions (for research purposes!) you can easily put them on as many computers as you want, without
breaking the bank, and when you move to another job, you won't find yourself crippled if they do not use your preferred software: just
bring yours along.


TIGR TM4 Microarray analysis suite.
Excellent suite of programs for everything from segementation and quantitation to cluster analysis, including a database (MySQL) to
organise your experiments.
I started using only their segmentation/quantitation program (Spotfinder), but in newer versions their analysis package (MIDAS) and
visualisation and exploration software (TMEV) have gotten much easier to use.

R/BioConductor/Limma
R is a very powerful statistical language. BioConductor is a collection of packages written in R for the analysis of microarray data. Limma
is a Bioconductor package for the
linear modelling or microarrays. It is command-line based, so not as straight forward to use as other
alternatives. But the flexibility is great. And it is really not very hard at all. There are good help pages, worked examples, and an online
BioConductor forum to help you get to grips with Limma. In addition, there is a LimmaGUI package that offers a more familiar windows-
like interface. LimmaGUI is not as flexible as Limma, but it is a lot easier to use, and it does allow you to incorporate your own scripts
written in R and customize the menus with your own scripts if you wish.


My favourite procedure
My usual way to analyse arrays at present go something like this:

Scanner - TIGR spotfinder (or GenePix) - Limma - webbased exploration (gene ontologies etc)
and
Scanner - TIGR spotfinder - TIGR MIDAS - TIGR TMEV

See the guide for details.


GEPAS
GEPAS is a collection of web-based tools that will allow you to do all kinds of analysis. Just scan your slides, quantitate, and feed the data
to one or more of the tools available. You can normalise, check quality, produce plots, perform clustering analysis, classify and compare
lists of genes, and much more.
The drawback is that is not as fast as running your own software on your own computer. But worth checking nevertheless!


Contact me.