The MAS 5.0 implementation is based upon availiable documentation. Clearly it might be impossible to completely duplicate Affymetrix, since there are pieces of their documentation that are left a little vague. Ben On Wed, 22 Jan 2003, Wilkinson, Mark wrote: > > This issue in general has kept me somewhat reluctant to use the affy > package. > > To the affy package authors: is the idea for affy to reproduce the > functionality of MAS 5.0 (at least for some parts)? In other words, should > I be able to do probe level summaries exactly like MAS5, or is affy's > version just a close approximation? > > > > Mark Wilkinson > Informatics Analyst > St. Jude Children's Research Hospital > Department of Pharmaceutical Sciences > > The opinions expressed here are my own and do not necessarily represent > those of St. Jude Children's Research Hospital. > > > -----Original Message----- > From: Helene Boucher [mailto:bouhel01@borabora.crchul.ulaval.ca] > Sent: Wednesday, January 22, 2003 10:09 AM > To: bioconductor@stat.math.ethz.ch > Subject: [BioC] MAS 5.0 > > > > I tried expresso to reproduce expression measures from MAS 5.0 but I get a > slight > difference from Affymetrix as illustrated on a set of 3 chips for selected > probe > sets. Any idea why that is? The version of Unix we have is SunOS Release > 5.8. > > Thanks, > > Helene > > > 100001_at 29.6222834908728 10.2880336221679 > 80.7973366171763 > 100002_at 9.94107964466495 62.9653009129815 > 53.7356175596969 > 100003_at 2262.85860839992 1539.08193903367 > 1910.0212276237 > 100004_at 305.100816246875 188.181043441888 > 239.902666074234 > 100005_at 434.482366404616 638.860562906946 > 589.86003783735 > 100006_at 61.9587790672468 32.9005081415387 > 61.5914892307955 > 100007_at 890.772109484179 988.46987085589 1076.17589775836 > 100009_r_at 89.4108950323307 46.1650654835421 > 54.0252748779586 > 100010_at 227.251880233698 401.164278682326 > 424.746403324436 > 100011_at 1157.66237359318 702.613181714498 > 882.033123210299 > 100012_at 893.256638440421 1511.96440725101 > 2054.18415149103 > -------------------------------------------------------------------- -------- > -------- > > > 100001_at 16 16 26.0 A 0.781017 > > > > 100002_at 16 16 10.2 A 0.910595 > > > > 100003_at 16 16 2271.8 P 0.000219 > > > > 100004_at 16 16 302.2 P 0.004863 > > > > 100005_at 16 16 439.1 P 0.006532 > > > > 100006_at 16 16 63.5 A 0.358690 > > > > 100007_at 16 16 894.7 P 0.001354 > > > > 100009_r_at 16 16 90.2 P 0.000805 > > > > 100010_at 16 16 227.4 P 0.009985 > -------------------------------------------------------------------- -------- > --------- > 100001_at 16 16 11.0 A 0.602006 > > > > 100002_at 16 16 58.8 A 0.438361 > > > > 100003_at 16 16 1545.5 P 0.000219 > > > > 100004_at 16 16 191.7 P 0.011447 > > > > 100005_at 16 16 641.4 P 0.002617 > > > > 100006_at 16 16 33.7 A 0.117160 > > > > 100007_at 16 16 992.9 P 0.001141 > > > > 100009_r_at 16 16 47.4 P 0.011447 > > > > 100010_at 16 16 402.6 P 0.001354 > -------------------------------------------------------------------- -------- > --------- > 100001_at 16 16 81.1 A 0.438361 > > > > 100002_at 16 16 44.4 A 0.581931 > > > > 100003_at 16 16 1917.2 P 0.000219 > > > > 100004_at 16 16 244.2 P 0.031336 > > > > 100005_at 16 16 591.0 P 0.001354 > > > > 100006_at 16 16 62.9 A 0.089405 > > > > 100007_at 16 16 1080.6 P 0.000673 > > > > 100009_r_at 16 16 54.5 P 0.011447 > > > > 100010_at 16 16 428.1 P 0.000959 > -------------------------------------------------------------------- -------- > -------- > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >

On Wed, Jan 22, 2003 at 12:58:00PM -0600, Wilkinson, Mark wrote: > > This issue in general has kept me somewhat reluctant to use the affy > package. > > To the affy package authors: is the idea for affy to reproduce the > functionality of MAS 5.0 (at least for some parts)? In other words, should > I be able to do probe level summaries exactly like MAS5, or is affy's > version just a close approximation? As stated, the implementation was done reading the Affymetrix "white paper" (available on the website of the company). The algorithms are doing what is claimed to be done (or at least what we understood to be done -- the source is open, we hope to correct errors as fast as they are reported ...). >From what I see in the data enclosed the difference between is roughly 1-2% between both implementations. I also see that the number from the MAS output are rounded, which would make one think that the differences have to do with truncation/rounding during the different steps. Effort was made to avoid that but the following may make you react. In 'pmcorrect.rma', I initially set the default parameter 'delta' to '2^-20' (as it is written in the specs, page 8). The default for delta is currently 2e-20 (I think the fix was suggested to obtain results similar to MAS 5.0, Rafael may want correct me on that). Are the specs or the code of MAS 5.0 right ? You are free to trust better MAS 5.0. Hopin' it helps, Laurent > > > > Mark Wilkinson > Informatics Analyst > St. Jude Children's Research Hospital > Department of Pharmaceutical Sciences > > The opinions expressed here are my own and do not necessarily represent > those of St. Jude Children's Research Hospital. > > > -----Original Message----- > From: Helene Boucher [mailto:bouhel01@borabora.crchul.ulaval.ca] > Sent: Wednesday, January 22, 2003 10:09 AM > To: bioconductor@stat.math.ethz.ch > Subject: [BioC] MAS 5.0 > > > > I tried expresso to reproduce expression measures from MAS 5.0 but I get a > slight > difference from Affymetrix as illustrated on a set of 3 chips for selected > probe > sets. Any idea why that is? The version of Unix we have is SunOS Release > 5.8. > > Thanks, > > Helene > > > 100001_at 29.6222834908728 10.2880336221679 > 80.7973366171763 > 100002_at 9.94107964466495 62.9653009129815 > 53.7356175596969 > 100003_at 2262.85860839992 1539.08193903367 > 1910.0212276237 > 100004_at 305.100816246875 188.181043441888 > 239.902666074234 > 100005_at 434.482366404616 638.860562906946 > 589.86003783735 > 100006_at 61.9587790672468 32.9005081415387 > 61.5914892307955 > 100007_at 890.772109484179 988.46987085589 1076.17589775836 > 100009_r_at 89.4108950323307 46.1650654835421 > 54.0252748779586 > 100010_at 227.251880233698 401.164278682326 > 424.746403324436 > 100011_at 1157.66237359318 702.613181714498 > 882.033123210299 > 100012_at 893.256638440421 1511.96440725101 > 2054.18415149103 > -------------------------------------------------------------------- -------- > -------- > > > 100001_at 16 16 26.0 A 0.781017 > > > > 100002_at 16 16 10.2 A 0.910595 > > > > 100003_at 16 16 2271.8 P 0.000219 > > > > 100004_at 16 16 302.2 P 0.004863 > > > > 100005_at 16 16 439.1 P 0.006532 > > > > 100006_at 16 16 63.5 A 0.358690 > > > > 100007_at 16 16 894.7 P 0.001354 > > > > 100009_r_at 16 16 90.2 P 0.000805 > > > > 100010_at 16 16 227.4 P 0.009985 > -------------------------------------------------------------------- -------- > --------- > 100001_at 16 16 11.0 A 0.602006 > > > > 100002_at 16 16 58.8 A 0.438361 > > > > 100003_at 16 16 1545.5 P 0.000219 > > > > 100004_at 16 16 191.7 P 0.011447 > > > > 100005_at 16 16 641.4 P 0.002617 > > > > 100006_at 16 16 33.7 A 0.117160 > > > > 100007_at 16 16 992.9 P 0.001141 > > > > 100009_r_at 16 16 47.4 P 0.011447 > > > > 100010_at 16 16 402.6 P 0.001354 > -------------------------------------------------------------------- -------- > --------- > 100001_at 16 16 81.1 A 0.438361 > > > > 100002_at 16 16 44.4 A 0.581931 > > > > 100003_at 16 16 1917.2 P 0.000219 > > > > 100004_at 16 16 244.2 P 0.031336 > > > > 100005_at 16 16 591.0 P 0.001354 > > > > 100006_at 16 16 62.9 A 0.089405 > > > > 100007_at 16 16 1080.6 P 0.000673 > > > > 100009_r_at 16 16 54.5 P 0.011447 > > > > 100010_at 16 16 428.1 P 0.000959 > -------------------------------------------------------------------- -------- > -------- > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor