Need help :D - Tukey Biweight anyone? (Oct/23/2008 )

Hi everyone

I have a little problem...and I hope someone can help.

We use R/Bioconductor to analyse our chip-chip experiments. All good there. It uses a Tukey Biweight Mean Algorithm in its calculations.
Now, my boss likes to see the actually probes in the UCSC browser, and so we have a Perl script that does that. Thing is, the Perl script does not use Tukey Biweight!! I want them both to be the same. My biggest problem is that I don't have a bioinformatician to help me (I am just a post-doc!). When I started this job I was told I would have help...but alas. I feel very lost in the world of bioinformatics and have sooooooooooooo much data I don't know what to do with it. Well, I know what I want to do with it, but don't know how to do it

Anyway, to get to the point..would anyone mind having a look at my Perl script? And posisbly tell me if it's easy to modify it so it uses Tukey?

I will love you forever

On another note: we really do want a bioinformatician so if anyone wants to collaborate???? I will PM you the script if you want to have a crack at it

Clare

-Clare-

Hi,

just have a look at the Ringo package in bioconductor. I think this package provide a perl script which provides your information.

additional information:
* http://www.biomedcentral.com/1471-2105/8/221
* www.bioconductor.org/packages/2.2/bioc/vignettes/Ringo/inst/doc/Ringo.pdf

BW

-xeroxed_yeti-

QUOTE (xeroxed_yeti @ Oct 23 2008, 04:11 PM)

Hi BW,
I had a look at the Ringo information and it says "The package’s scripts directory also contains Perl scripts that allow the conversion of multiple output files from common alignment tools such as Exonerate into one file that corresponds to a POS file".
Now the question is...how to find the scripts directory!?

Clare

PS: It's a bit creepy that you can display my IP address and where I'm working from

-Clare-

What kind of machine are you using? Linux/Mac/Windows (which one). Did you do a standard install? If you want help with your perl script post it.

The IP etc is all very simple to do - all the information is passed by your browser to each page you visit - unless you use proxies.

-perlmunky-

Hey Clara,

as perlmunky said, we need more input

by the way, BW means best wishes

-xeroxed_yeti-

QUOTE (perlmunky @ Oct 23 2008, 11:53 PM)

Hey there

I am using Windows and I didn't install anything because we're not allowed

So IT did it. I will post you the perl script when I get back to work on MOnday

wOOt!

Happy Friday!!!

Clare

-Clare-

QUOTE (xeroxed_yeti @ Oct 24 2008, 08:41 AM)

Hey Clara,

as perlmunky said, we need more input

by the way, BW means best wishes

hehehe. I am a muppet!
BTW, It's ClarE not ClarA

Clare

-Clare-

The perl code is yuck!

Have you seen this: http://rss.acs.unt.edu/Rdoc/library/affy/h...y.biweight.html

CODE

tukey.biweight <- function(x, c=5, epsilon=0.0001)
  {
    m <- median(x)
    s <- median(abs(x - m))
    u <- (x - m) / (c * s + epsilon)
    w <- rep(0, length(x))
    i <- abs(u) <= 1
    w[i] <- ((1 - u^2)^2)[i]
    t.bi <- sum(w * x) / sum(w)
    return(t.bi)
  }

tukeybiweight <-  function(x, c=5, epsilon=0.0001)
  list(exprs=apply(x,2,tukey.biweight,c=c,epsilon=epsilon),se.exprs=rep(NA,ncol(x)))

-perlmunky-

The perl code I sent you?

I didn't write it - the 'Bioinformatician' in our department did. hehehe

The one-step Tukey looks awesome but that's for R isn't it? Can you get an output in R that allows you to visualise probes on the UCSC browser?

Clare

-Clare-

Yup, the code you sent me

I really haven't had a chance to spend that much time with it yet ... I seem to have things to do for a change! I will probably have another look at it over the weekend. I would imagine that you can take the R output at chuck it at UCSC tracks...

-perlmunky-