[BioC] Re: Additional questions regarding LIMMA

[Gordon Smyth]

At 08:26 PM 18/06/2003, muehleisen wrote:
>Dear Dr. Smyth,
>
>I beg your pardon that I  send you additional questions regarding LIMMA 
>but it's very important for my data analysis!
>
>On the page "Print-Order Normalization - Methods" (html guide), you wrote 
>that the printorder is assumed to start in the top right hand corner of 
>each tip-group and proceeds left and down by rows. However, this is true 
>for your array layout but not for mine. Our arrayer starts printing in the 
>top left hand corner. I read that the function "printorder" includes the 
>argument "start" by which one can determine a certain printorder. But, I 
>wasn't able to find that this argument is defined for the functions 
>"normalizeforPrintorder" and "plotPrintorder". May I ask you why?

Well, because our arrayer prints from "topright" and I didn't have time to 
write the function more generally. I have now added support for printing 
from "topleft".

>Another question concerns the "PlotPrintorder" argument "method". I tried 
>six different cases using "normalizeForPrintorder". At first, I didn't 
>state the "method" argument explicitly. At second, I explicitly wrote 
>method="loess" and span=0.1. And, at third, I wrote method="none". Each 
>case was carried out using either "separate.channels=TRUE" or 
>"separate.channels=FALSE". I compared the corresponding plots intensively 
>but I couldn't find striking differences between the cases argument 
>"method" left out, argument "method="loess" and span=0.1" and argument 
>"method="none".  If I leave out the argument "method" does the function 
>"PlotPrintorder" set "method="loess"" and "span" by default? In addition, 
>is "method="none" defined for "PlotPrintorder" as it is the case for 
>"normalizeWithinArrays"?

No, "none" is not a defined choice. The help file tells you that the 
choices for method are "loess" and "plate".

I usually write my functions so that any unrecognized method is equivalent 
to the default, in this case "loess". Perhaps it would be better to give an 
explicit error when an unrecognized method is asked for - I have added in a 
line to achieve that treatment now.

Gordon

>Sincerely,
>
>Thomas W. Muehleisen
>
>Max Planck Institute for Brain Research
>Deutschordenstrasse 46
>D-60528 Frankfurt/Main