Using RUVs on Deseq2 but clustering still seems wrong
1
0
Entering edit mode
r.tom • 0
@rtom-22042
Last seen 5.2 years ago
UK

Hello! Apologies if this question has been asked before. I am not very experienced in bioinformatics... I am analysing 15 samples, which include 3 biological replicates of the same cell line (A, B, C) and 5 different treatments (1,2,3,4,5). I am interested in the effect of treatment 1 (ctl) versus each of the other treatments. However, once I import my data into Deseq2 file and plot PCA, I can see the samples cluster by biological replicate rather than by treatment (even if I have both these factors in the degsign), so I wonder if this reflects a batch effect of the experiment itself. I tried using RUVs with k=3 (after several trials) and the samples cluster quite nicely by treatment. However when I apply these variants (W1 + W2 + W3) to the design and repeat the PCA the clustering in the PCA doesn't change much, neither does the heatmap of the most variable genes.. am I doing something wrong? Where/how will I see any effect of the RUV normalisation? What is the most appropriate way to show clustering if there is such a clear replicate bias? Thank you very much for your help! I really appreciate any feedback.

samples <- read.table(file.path(dir,"samplestest.txt"), header=TRUE, stringsAsFactors = TRUE) rownames(samples) <- samples$sample files <- file.path(dir,"salmon", paste(samples$sample, ".txt", sep='') ) names(files) <- samples$samples txi <- tximport(files, type="salmon", tx2gene=tx2gene, ignoreTxVersion = TRUE) reading in files with readtsv 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 transcripts missing from tx2gene: 13365 summarizing abundance summarizing counts summarizing length ddsTxi <- DESeqDataSetFromTximport(txi,colData = samples, design = ~ replicate + treatment) using counts and average transcript lengths from tximport dds <- DESeq(ddsTxi) estimating size factors using 'avgTxLength' from assays(dds), correcting for library size estimating dispersions gene-wise dispersion estimates mean-dispersion relationship final dispersion estimates fitting model and testing dds <- dds[ rowSums(counts(dds)) > 10, ] rld <- rlog(dds, blind=FALSE) plotPCA(rld, intgroup=c("replicate")) PCA

counts <- counts(dds, normalized=FALSE) set <- newSeqExpressionSet(counts = counts, phenoData = samples) sets <- RUVs(set, unique(rownames(set)), k=3, differences) library(RColorBrewer) colors <- brewer.pal(3, "Set2") par(mfrow = c(1,2)) plotPCA(set, col=colors[samples$treatment], cex=1.2, main = 'without RUVs') plotPCA(sets, col=colors[samples$treatment], cex=1.2, main = 'with RUVs') PCA plot RUVs

ddsTxiruv <- DESeqDataSetFromTximport(txi, + colData = pData(sets), + design = ~ W1 + W2 + W_3 + treatment) using counts and average transcript lengths from tximport

ddsruv <- DESeq(ddsTxiruv) estimating size factors using 'avgTxLength' from assays(dds), correcting for library size estimating dispersions gene-wise dispersion estimates mean-dispersion relationship final dispersion estimates fitting model and testing 104 rows did not converge in beta, labelled in mcols(object)$betaConv. Use larger maxit argument with nbinomWaldTest nc <- counts(ddsruv, normalized=TRUE) filter <- rowSums(nc >= 50) >= 15 ddsruv <- ddsruv[filter,] ddsruv <- DESeq(ddsruv) using pre-existing normalization factors estimating dispersions found already estimated dispersions, replacing these gene-wise dispersion estimates mean-dispersion relationship final dispersion estimates fitting model and testing rldruv <- rlog(ddsruv, blind=FALSE) plotPCA(rldruv, intgroup=c("replicate")) PCA after RUV

topVarGenes <- head(order(rowVars(assay(rld)), decreasing = TRUE), 10) mat <- assay(rld)[ topVarGenes, ] anno <- as.data.frame(colData(rld)[, c("treatment", "replicate")]) pheatmap(mat, annotationcol = anno, showrownames = F)

Heatmap rld

topVarGenes <- head(order(rowVars(assay(rldruv)), decreasing = TRUE), 10) mat <- assay(rldruv)[ topVarGenes, ] anno <- as.data.frame(colData(rldruv)[, c("treatment", "replicate")]) pheatmap(mat, annotationcol = anno, show_rownames = F)

Heatmap rld_ruv

sessionInfo(package=NULL) R version 3.6.0 (2019-04-26) Platform: x86_64-apple-darwin15.6.0 (64-bit) Running under: macOS Mojave 10.14.6

Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib

locale: [1] enGB.UTF-8/enGB.UTF-8/enGB.UTF-8/C/enGB.UTF-8/en_GB.UTF-8

attached base packages: [1] parallel stats4 stats graphics grDevices utils datasets methods
[9] base

other attached packages: [1] sva3.32.1 genefilter1.66.0 mgcv1.8-29
[4] nlme
3.1-141 RUVSeq1.18.0 edgeR3.26.8
[7] limma3.40.6 EDASeq2.18.0 ShortRead1.42.0
[10] GenomicAlignments
1.20.1 Rsamtools2.0.1 Biostrings2.52.0
[13] XVector0.24.0 RColorBrewer1.1-2 biomaRt2.40.4
[16] pheatmap
1.0.12 tximportData1.12.0 readr1.3.1
[19] tximport1.12.3 DESeq21.24.0 SummarizedExperiment1.14.1 [22] DelayedArray0.10.0 BiocParallel1.18.1 matrixStats0.55.0
[25] Biobase2.44.0 GenomicRanges1.36.1 GenomeInfoDb1.20.0
[28] IRanges
2.18.3 S4Vectors0.22.1 BiocGenerics0.30.0

loaded via a namespace (and not attached): [1] bitops1.0-6 bit640.9-7 progress1.2.2
[4] httr
1.4.1 tools3.6.0 backports1.1.4
[7] R62.4.0 rpart4.1-15 Hmisc4.2-0
[10] DBI
1.0.0 lazyeval0.2.2 colorspace1.4-1
[13] nnet7.3-12 gridExtra2.3 prettyunits1.0.2
[16] curl
4.2 bit1.1-14 compiler3.6.0
[19] htmlTable1.13.2 labeling0.3 rtracklayer1.44.4
[22] scales
1.0.0 checkmate1.9.4 DESeq1.36.0
[25] stringr1.4.0 digest0.6.21 foreign0.8-72
[28] R.utils
2.9.0 base64enc0.1-3 pkgconfig2.0.3
[31] htmltools0.3.6 htmlwidgets1.3 rlang0.4.0
[34] rstudioapi
0.10 RSQLite2.1.2 jsonlite1.6
[37] hwriter1.3.2 acepack1.4.1 R.oo1.22.0
[40] RCurl
1.95-4.12 magrittr1.5 GenomeInfoDbData1.2.1 [43] Formula1.2-3 Matrix1.2-17 Rcpp1.0.2
[46] munsell
0.5.0 R.methodsS31.7.1 stringi1.4.3
[49] MASS7.3-51.4 zlibbioc1.30.0 grid3.6.0
[52] blob
1.2.0 crayon1.3.4 lattice0.20-38
[55] splines3.6.0 GenomicFeatures1.36.4 annotate1.62.0
[58] hms
0.5.1 locfit1.5-9.1 zeallot0.1.0
[61] knitr1.25 pillar1.4.2 geneplotter1.62.0
[64] XML
3.98-1.20 latticeExtra0.6-28 BiocManager1.30.4
[67] data.table1.12.2 vctrs0.2.0 gtable0.3.0
[70] assertthat
0.2.1 ggplot23.2.1 xfun0.9
[73] aroma.light3.14.0 xtable1.8-4 survival2.44-1.1
[76] tibble
2.1.3 AnnotationDbi1.46.1 memoise1.1.0
[79] cluster_2.1.0

deseq2 ruv batch • 1.4k views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 6 days ago
United States

This is in fact asked a lot. It is also a Frequently Asked Question in the vignette.

ADD COMMENT

Login before adding your answer.

Traffic: 326 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6