Hi, I'm trying to analyze the differential expression between control and patient groups in a microarray from Illumina HumanHT12_V4. I would like to know what advantages or disadvantages have the use of the "detectionCall" function from "lumi" package. Once I've removed outliers and normalized, I've tried to reduce the number of genes with "detectionCall" in order to filter possible false positives. Then, I obtain the list of differentially expressed genes by applying the "limma" package ("lmFit" and "eBayes" functions). However, that list usually includes a list with more differentially expressed genes when I'm using the "detectionCall" function. Is this usual? If I've reduced the number of false positive genes, how is it possible that I obtain a higher list? Is my interpretation of "detectionCall" correct? Thanks in advance, Francisco. -- output of sessionInfo(): R version 2.15.0 (2012-03-30) Platform: x86_64-pc-mingw32/x64 (64-bit) locale: [1] LC_COLLATE=Spanish_Spain.1252 LC_CTYPE=Spanish_Spain.1252 [3] LC_MONETARY=Spanish_Spain.1252 LC_NUMERIC=C [5] LC_TIME=Spanish_Spain.1252 attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] lumiHumanIDMapping_1.10.0 arrayQualityMetrics_3.12.0 [3] lumiHumanAll.db_1.18.0 org.Hs.eg.db_2.7.1 [5] RSQLite_0.11.3 DBI_0.2-6 [7] annotate_1.34.1 AnnotationDbi_1.18.4 [9] lumi_2.8.0 nleqslv_2.0 [11] methylumi_2.2.0 ggplot2_0.9.3.1 [13] reshape2_1.2.2 scales_0.2.3 [15] Biobase_2.16.0 BiocGenerics_0.2.0 [17] limma_3.12.3 loaded via a namespace (and not attached): [1] affy_1.34.0 affyio_1.24.0 affyPLM_1.32.0 [4] beadarray_2.6.0 BeadDataPackR_1.8.0 bigmemory_4.2.11 [7] BiocInstaller_1.4.9 Biostrings_2.24.1 bitops_1.0-5 [10] BSgenome_1.24.0 Cairo_1.5-2 cluster_1.14.4 [13] colorspace_1.2-2 dichromat_2.0-0 digest_0.6.3 [16] DNAcopy_1.30.0 genefilter_1.38.0 GenomicRanges_1.8.13 [19] genoset_1.6.0 grid_2.15.0 gtable_0.1.2 [22] hdrcde_2.16 Hmisc_3.10-1 hwriter_1.3 [25] IRanges_1.14.4 KernSmooth_2.23-10 labeling_0.1 [28] lattice_0.20-6 latticeExtra_0.6-24 MASS_7.3-23 [31] Matrix_1.0-12 mgcv_1.7-22 munsell_0.4 [34] nlme_3.1-108 plyr_1.8 preprocessCore_1.18.0 [37] proto_0.3-10 RColorBrewer_1.0-5 RCurl_1.95-4.1 [40] Rsamtools_1.8.6 rtracklayer_1.16.3 setRNG_2011.11-2 [43] splines_2.15.0 stats4_2.15.0 stringr_0.6.2 [46] survival_2.37-4 SVGAnnotation_0.93-1 tools_2.15.0 [49] vsn_3.24.0 XML_3.96-1.1 xtable_1.7-1 [52] zlibbioc_1.2.0 -- Sent via the guest posting facility at bioconductor.org.

Hi Francisco I believe this is because limma needs to estimate the background variation. The filtered unexpressed genes have lower variation. After you filtered them, the estimated background variation will be higher, which may result in a shorter list of differentially expressed genes. You can check the limma vignette for more details. Pan On Thu, Jun 6, 2013 at 12:38 AM, Francisco Ortuno [guest] < guest@bioconductor.org> wrote: > > Hi, > > I'm trying to analyze the differential expression between control and > patient groups in a microarray from Illumina HumanHT12_V4. I would like to > know what advantages or disadvantages have the use of the "detectionCall" > function from "lumi" package. > > Once I've removed outliers and normalized, I've tried to reduce the number > of genes with "detectionCall" in order to filter possible false positives. > Then, I obtain the list of differentially expressed genes by applying the > "limma" package ("lmFit" and "eBayes" functions). However, that list > usually includes a list with more differentially expressed genes when I'm > using the "detectionCall" function. Is this usual? If I've reduced the > number of false positive genes, how is it possible that I obtain a higher > list? Is my interpretation of "detectionCall" correct? > > Thanks in advance, > Francisco. > > > -- output of sessionInfo(): > > R version 2.15.0 (2012-03-30) > Platform: x86_64-pc-mingw32/x64 (64-bit) > > locale: > [1] LC_COLLATE=Spanish_Spain.1252 LC_CTYPE=Spanish_Spain.1252 > [3] LC_MONETARY=Spanish_Spain.1252 LC_NUMERIC=C > [5] LC_TIME=Spanish_Spain.1252 > > attached base packages: > [1] stats graphics grDevices utils datasets methods base > > other attached packages: > [1] lumiHumanIDMapping_1.10.0 arrayQualityMetrics_3.12.0 > [3] lumiHumanAll.db_1.18.0 org.Hs.eg.db_2.7.1 > [5] RSQLite_0.11.3 DBI_0.2-6 > [7] annotate_1.34.1 AnnotationDbi_1.18.4 > [9] lumi_2.8.0 nleqslv_2.0 > [11] methylumi_2.2.0 ggplot2_0.9.3.1 > [13] reshape2_1.2.2 scales_0.2.3 > [15] Biobase_2.16.0 BiocGenerics_0.2.0 > [17] limma_3.12.3 > > loaded via a namespace (and not attached): > [1] affy_1.34.0 affyio_1.24.0 affyPLM_1.32.0 > [4] beadarray_2.6.0 BeadDataPackR_1.8.0 bigmemory_4.2.11 > [7] BiocInstaller_1.4.9 Biostrings_2.24.1 bitops_1.0-5 > [10] BSgenome_1.24.0 Cairo_1.5-2 cluster_1.14.4 > [13] colorspace_1.2-2 dichromat_2.0-0 digest_0.6.3 > [16] DNAcopy_1.30.0 genefilter_1.38.0 GenomicRanges_1.8.13 > [19] genoset_1.6.0 grid_2.15.0 gtable_0.1.2 > [22] hdrcde_2.16 Hmisc_3.10-1 hwriter_1.3 > [25] IRanges_1.14.4 KernSmooth_2.23-10 labeling_0.1 > [28] lattice_0.20-6 latticeExtra_0.6-24 MASS_7.3-23 > [31] Matrix_1.0-12 mgcv_1.7-22 munsell_0.4 > [34] nlme_3.1-108 plyr_1.8 preprocessCore_1.18.0 > [37] proto_0.3-10 RColorBrewer_1.0-5 RCurl_1.95-4.1 > [40] Rsamtools_1.8.6 rtracklayer_1.16.3 setRNG_2011.11-2 > [43] splines_2.15.0 stats4_2.15.0 stringr_0.6.2 > [46] survival_2.37-4 SVGAnnotation_0.93-1 tools_2.15.0 > [49] vsn_3.24.0 XML_3.96-1.1 xtable_1.7-1 > [52] zlibbioc_1.2.0 > > -- > Sent via the guest posting facility at bioconductor.org. > [[alternative HTML version deleted]]