Dear all, Thanks to Mike Kuziora (Genelogic) the mas5calls function in affy now gives results almost identical to Affymetrix MAS 5.0. This is in the latest release 1.3.28 Thanks to Crispin Miller the devel pacakge of affy 1.4.13 has a version of the mas5calls that works much faster. the results are almost identical to those in 1.3.28. -r

I had a question relating to comparing a couple different normalizations (like mas5 vs rma). If I used the default affy software, it produces P,M,A values for each (as does mas5value call) expression measurement. But what is a good method to reject measurements after rma is called since that doesn't produce the P,M,A values? What do people genrally do? Tom Fahland -----Original Message----- From: Rafael A. Irizarry [mailto:ririzarr@jhsph.edu] Sent: Thursday, February 05, 2004 21:42 To: bioconductor@stat.math.ethz.ch Subject: [BioC] mas5calls now almost like affymetrix's Dear all, Thanks to Mike Kuziora (Genelogic) the mas5calls function in affy now gives results almost identical to Affymetrix MAS 5.0. This is in the latest release 1.3.28 Thanks to Crispin Miller the devel pacakge of affy 1.4.13 has a version of the mas5calls that works much faster. the results are almost identical to those in 1.3.28. -r _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor

The P, M, A calls from MAS5 don't (IMO) tell you whether the gene is being expressed or not, but how well the PM signal can be distinguished from the IM signal. Since rma doesn't use the MM or IM signal as an estimator of non-specific binding, there is no way to make this call. I think you could make a very strong case that you shouldn't actually reject any measurements based on the expression values anyway. The gene may not be expressed in some of your samples, but expressed at a very low level in other samples. I would argue that these genes are likely very interesting. Also, the level of expression does not a priori give you an idea of the phenotypic effect. In fact, for many genes I would imagine low expression is all you ever get. For instance, gene expression in the brain is extremely difficult to measure because the amount of mRNA produced is so low. It seems to me that most of the genes that are at the signal/noise boundary are lost when you start doing statistical tests anyway, so there is no profit in eliminating them beforehand. Best, Jim James W. MacDonald Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623 >>> "Tom R. Fahland" <tfahland@genomatica.com> 02/06/04 11:58AM >>> I had a question relating to comparing a couple different normalizations (like mas5 vs rma). If I used the default affy software, it produces P,M,A values for each (as does mas5value call) expression measurement. But what is a good method to reject measurements after rma is called since that doesn't produce the P,M,A values? What do people genrally do? Tom Fahland -----Original Message----- From: Rafael A. Irizarry [mailto:ririzarr@jhsph.edu] Sent: Thursday, February 05, 2004 21:42 To: bioconductor@stat.math.ethz.ch Subject: [BioC] mas5calls now almost like affymetrix's Dear all, Thanks to Mike Kuziora (Genelogic) the mas5calls function in affy now gives results almost identical to Affymetrix MAS 5.0. This is in the latest release 1.3.28 Thanks to Crispin Miller the devel pacakge of affy 1.4.13 has a version of the mas5calls that works much faster. the results are almost identical to those in 1.3.28. -r _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor