[BioC] Error doing Mas5 calculation with Simpleaffy package

Crispin Miller CMiller at PICR.man.ac.uk
Wed May 26 10:59:06 CEST 2004

Thanks Rudy,
I'll have a look into things and produce an update as soon as I can,
hopefully in the next day or two...
Whilst I'm at it, I'll add the 4302 chips into the QC stuff...

In the meantime, I strongly suggest treating the mas5 calls in
simpleaffy with caution. :-( 

-----Original Message-----
From: Rudy.vanEijsden at GEN.unimaas.nl
[mailto:Rudy.vanEijsden at GEN.unimaas.nl] 
Sent: 25 May 2004 15:16
To: Crispin Miller
Cc: bioconductor at stat.math.ethz.ch
Subject: RE: [BioC] Error doing Mas5 calculation with Simpleaffy package

Hi Crispin,

We have also looked into the C-code (simpleaffy v1.09) and found the
following errors:
(1) On line 547:
According to us the chip is being remapped onto a grid. Hence, if the
number of columns or rows of the affy-chip respectively devided by the
number of columns or rows of the grid does not yield an integer, then
the number of rows/columns to be places in each cell of the grid was
rounded down instead of up. This resulted in memory crashes.

(2) On lines 515 and 602:
According to us the index for the raw and corrected intensities are
wrong as c-code start counting at 0 and not 1, therefore the index needs
to be decreased by 1.

(3) On lines 506 and 507:
According to us all the variable related to "x" and "y" respectively
correspond to columns and rows. Therefore if the affy-chip is not square
these lines are wrong as "nrow" and "ncol" don't respectively correspond
to "y" and "x".

Attached you will find a patch for the "simpleaffy2.c" source file.

After these corrections we have compared the MAS5 expression values of 4
chips resulting from the calculations of the simpleaffy v1.0 and
simpleaffy v1.09. The results were quite comparable, except for 54
values that differed by more than 0.1 units (on log scale) and ranged
from -1.635837 to 1.448547 units.


Rudy van Eijsden.

Maastricht University
Cluster of Genetics and Cell Biology
Department of Clinical Genetics

Visiting Address
Universiteitssingel 50, Room No 5.140
6229 ER Maastricht
The Netherlands

Postal Address
P.O. box 616, post-box 16
6200 MD Maastricht
The Netherlands

+31 (0)43 3881916 (Office)
+31 (0)43 3881995 (Secretariat)

+31 (0)43 3884573

E-mail & Internet
rudy.vaneijsden at gen.unimaas.nl
http://www.gen.unimaas.nl <http://www.gen.unimaas.nl/>  

-----Original Message-----
From: Crispin Miller [mailto:CMiller at PICR.man.ac.uk]
Sent: Tuesday, May 11, 2004 12:28 PM
To: Rudy.vanEijsden at gen.unimaas.nl
Cc: bioconductor at stat.math.ethz.ch
Subject: RE: [BioC] Error doing Mas5 calculation with Simpleaffy package

Hi Rudy,
Sounds like a bug in the C library somewhere... I'm just pulling down
some yeast chips to play with - I'll get back to you when I've
identified the problem... Cheers, Crispin

-----Original Message-----
From: bioconductor-bounces at stat.math.ethz.ch
[mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of
Rudy.vanEijsden at gen.unimaas.nl
Sent: 07 May 2004 13:37
To: bioconductor at stat.math.ethz.ch
Subject: [BioC] Error doing Mas5 calculation with Simpleaffy package

Dear all,

I get the following error when doing the Mas5 calculation using the
Simpleaffy package:

Rgui.exe has generated errors and will be closed by Windows. You will
need to restart the program.

An error log is being created.

More details:
R-version: 1090 for Windows.
Simpleaffy version: Version: 1.09, Packaged: Fri Apr 16 03:23:08 2004;
Chips: 4 affymetrix yeast chips (YG-S98)

Code used:
>raw.data <- read.affy()
>data.mas <- call.exprs(raw.data, "mas5")

This is the moment when the error pups up.

Using rma in stead of mas5 does not give an error (>data.rma <-
call.exprs(raw.data, "rma")

Do you have any idea how to solve this problem.

I have tried doing the mas5 calculation using the affy package....and
this works fine. But I would like to use the simpleaffy package.



rudy.vaneijsden at gen.unimaas.nl

Bioconductor mailing list
Bioconductor at stat.math.ethz.ch

This email is confidential and intended solely for the use o...{{dropped}}

More information about the Bioconductor mailing list