I'm having troubles with some new arrays we're working with. After some effort, I've managed to read .csv output from PerkinElmer ScanArray into an RGList object. Warning to anyone else -- the .gpr files this program generates do not conform strictly to the gpr file format, and will make both limma and marray very unhappy. I have a gal file which appears to correctly identify block, column, row, ID, and name, and have even managed to make a SpotTypes file that shows me the control probes. The RGList looks like this (sorry for the large message): > RG2 An object of class "RGList" $R slide_13295067 slide_13295072 slide_13295071 slide_13295073 slide_13295075 slide_13295079 slide_13295080 slide_13295203 slide_13295068 slide_13295214 [1,] 61323 57928 57404 55737 61890 58153 63823 53921 57871 59289 [2,] 57955 51322 55691 60830 48322 64418 59731 47373 53971 54898 [3,] 1057 3361 1050 1097 1651 1174 1349 2176 1220 1471 [4,] 3185 7716 5754 5925 5850 8287 8498 6246 3255 7011 [5,] 1306 810 1069 533 978 1485 277 326 1190 679 slide_13295879 slide_13295887 [1,] 58546 58522 [2,] 59005 61561 [3,] 1440 1566 [4,] 13061 12547 [5,] 24787 648 38972 more rows ... $G slide_13295067 slide_13295072 slide_13295071 slide_13295073 slide_13295075 slide_13295079 slide_13295080 slide_13295203 slide_13295068 slide_13295214 [1,] 59066 58672 47495 47399 63671 59502 53084 62982 55238 52893 [2,] 52584 54957 39850 50818 54542 62064 47929 57410 48840 49985 [3,] 2443 1444 812 2001 598 1127 2465 1558 1308 1148 [4,] 5249 6511 2926 7372 4312 5657 10363 8768 2464 6809 [5,] 1662 1060 481 954 936 1071 608 920 512 438 slide_13295879 slide_13295887 [1,] 57486 51338 [2,] 58792 56861 [3,] 959 1349 [4,] 12943 9490 [5,] 776 994 38972 more rows ... $Rb slide_13295067 slide_13295072 slide_13295071 slide_13295073 slide_13295075 slide_13295079 slide_13295080 slide_13295203 slide_13295068 slide_13295214 [1,] 172 193 186 208 220 174 205 237 183 185 [2,] 168 191 183 252 197 214 195 204 202 259 [3,] 140 151 145 150 157 146 151 185 142 215 [4,] 148 144 155 153 147 152 149 190 152 182 [5,] 133 140 147 147 132 143 147 185 144 147 slide_13295879 slide_13295887 [1,] 208 228 [2,] 210 230 [3,] 171 152 [4,] 147 143 [5,] 152 145 38972 more rows ... $Gb slide_13295067 slide_13295072 slide_13295071 slide_13295073 slide_13295075 slide_13295079 slide_13295080 slide_13295203 slide_13295068 slide_13295214 [1,] 252 250 261 319 301 285 355 588 301 319 [2,] 247 273 280 300 279 294 324 568 251 310 [3,] 190 197 208 246 219 187 265 497 223 249 [4,] 185 198 219 247 244 179 263 473 217 239 [5,] 177 206 210 242 278 196 248 501 233 247 slide_13295879 slide_13295887 [1,] 310 270 [2,] 287 286 [3,] 237 197 [4,] 223 222 [5,] 233 226 38972 more rows ... $targets [1] "slide_13295067.csv" "slide_13295072.csv" "slide_13295071.csv" "slide_13295073.csv" "slide_13295075.csv" 7 more rows ... $genes Block Column Row ID Name Status 1 1 1 1 mCP000073 gi|21070949|ref|NM_019639.2|_151 control 2 1 2 1 mCP000085 gi|21070949|ref|NM_019639.2|_151 control 3 1 3 1 mCT000169 NM_008512.90|chr10|-|127557334|127558231_72 control 4 1 4 1 mCP000181 gi|31981889|ref|NM_009735.2|_43 control 5 1 5 1 mCT000265 NM_011701.9|chr2|+|13576298|13576633_142 control 38971 more rows ... $printer $ngrid.r [1] 12 $ngrid.c [1] 4 $nspot.r [1] 28 $nspot.c [1] 29 attr(,"class") [1] "PrintLayout" Trying to normalize with either loess or robustspline yields: > MA <- normalizeWithinArrays(RG2) Error in switch(method, loess = { : printer layout information does not match M row dimension > MA <- normalizeWithinArrays(RG2, method="robustspline") Loading required package: MASS Loading required package: splines Error in normalizeRobustSpline(object$M[, j], object$A[, j], layout, df = df, : (subscript) logical subscript too long Both suggest that something is amiss with my layout information, but I cannot begin to imagine how to diagnose the exact problem. Help! Thanks, Anand C. Patel, MD Washington University School of Medicine acpatel at usa.net

A quick look at the CSV files generated by PerkinElmer's ScanArray finds a bottom row that only has in it "END DATA". Thus, as Dr. Smyth predicted: > RG2$R[38977,] slide_13295067 slide_13295072 slide_13295071 slide_13295073 slide_13295075 slide_13295079 slide_13295080 slide_13295203 slide_13295068 slide_13295214 NA NA NA NA NA NA NA NA NA NA slide_13295879 slide_13295887 NA NA > Are we the only people using PerkinElmer's ScanArray software? I did my due diligence in terms of searching the list archives prior to posting, and didn't see any other posts. It is frustrating that even the "standard" gpr format would not work, and that the csv format output is nonstandard. However, it doesn't seem like it should be terribly difficult to modify the genepix routine in read.maimages into a "scanarray_gpr" option, much as was done with bluefuse. Is this something I should work through and submit? Also, besides the obvious step of going through the CSV files and erasing the last row, is there an easy way to remove row 38977 from the RGlist as a whole? Much thanks to Dr. Smyth for both limma and his assistance. Thanks, Anand C. Patel, MD Washington University School of Medicine acpatel at usa.net >[BioC] problems normalizing in limma >Gordon Smyth smyth at wehi.edu.au >Thu Oct 13 12:25:57 CEST 2005 > >Previous message: [BioC] annotations for Codelink arrays >Next message: [BioC] Source Code of affy package >Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] > >The output you give shows that all your intensity matrices (R, G, Rb and >Gb) are one row longer than your annotation data.frame RG2$genes. This >cannot be. The layout information agree with the dimension of RG2 $genes. > >My guess is that your input gpr files contained a spurious extra row which >is not real data. If so, you need to fix them to remove this spurious row, >or simply remove the last row of all the intensity columns. (Check first >that RG2$R[38977,] is not real data.) > >Gordon