Hi, I have a RNA-Seq dataset produced on a Illumina GA and wonder how to account for both technical and biological replication using edgeR. The dataset consist of 48 individual samples (time series with 4-6 biological replica per time point, 9 time points in total) which have been sequenced in three GA runs (3 technical replica) using 12-plex sequencing (4 lanes/run) - that is a total of 144 profiles representing 3 technical replica and 4-6 biological replica. Simply adding the counts from the 3 tech rep forming 48 profiles would be a start point, but it would be more powerful to include the tech rep in the analysis as well... Best regards, Jakob ############################### R version 2.11.0 (2010-04-22) x86_64-pc-mingw32 locale: [1] LC_COLLATE=Danish_Denmark.1252 LC_CTYPE=Danish_Denmark.1252 LC_MONETARY=Danish_Denmark.1252 LC_NUMERIC=C [5] LC_TIME=Danish_Denmark.1252 attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] edgeR_1.6.0 loaded via a namespace (and not attached): [1] limma_3.4.0 tools_2.11.0

Hi Jacob Technical replicates exists on many levels. My guess from your description is that the technical replicates you have is the same library prep injected into different lanes of the same (or different flowcells). It has been well established that technical replication at this level is close to poisson and that the lane and flowcell effect is pretty minor (See for example Bullard et al (2010) BMC Bioinformatics). So you will not loose much by aggregating and you will certainly gain a lot in reduced complexity. It is hard enough to suitably account for biological variation that I would start with that. And I would also recommend you to have a look at DESeq as well (which currently only does two group comparisons). Indeed, one can argue that the use of several lanes is not so much a matter of replication, but rather a consequence of aiming for greater sequencing depth. Things might be different if you have separate library preps. Kasper On Mon, Apr 26, 2010 at 9:32 AM, Jakob Hedegaard <jakob.hedegaard at="" agrsci.dk=""> wrote: > Hi, > I have a RNA-Seq dataset produced on a Illumina GA and wonder how to account for both technical and biological replication using edgeR. > The dataset consist of 48 individual samples (time series with 4-6 biological replica per time point, 9 time points in total) which have been sequenced in three GA runs (3 technical replica) using 12-plex sequencing (4 lanes/run) - that is a total of 144 profiles representing 3 technical replica and 4-6 biological replica. > Simply adding the counts from the 3 tech rep forming 48 profiles would be a start point, but it would be more powerful to include the tech rep in the analysis as well... > > Best regards, > Jakob > > > ############################### > R version 2.11.0 (2010-04-22) > x86_64-pc-mingw32 > > locale: > [1] LC_COLLATE=Danish_Denmark.1252 ?LC_CTYPE=Danish_Denmark.1252 ? ?LC_MONETARY=Danish_Denmark.1252 LC_NUMERIC=C > [5] LC_TIME=Danish_Denmark.1252 > > attached base packages: > [1] stats ? ? graphics ?grDevices utils ? ? datasets ?methods ? base > > other attached packages: > [1] edgeR_1.6.0 > > loaded via a namespace (and not attached): > [1] limma_3.4.0 ?tools_2.11.0 > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor >

Dear Jakob, Yes, pool the technical replicates. Unlike microarrays, there's no gain from keeping the tech reps separate, unless you have good evidence they show greater than Poisson variability. You could fit common dispersion to the technical replicates to check this, if you feel the need. (To do that, setup one grouping factor which takes the same level for each run of technical replicates.) Best wishes Gordon > Date: Mon, 26 Apr 2010 15:32:25 +0200 > From: Jakob Hedegaard <jakob.hedegaard at="" agrsci.dk=""> > To: "bioconductor at stat.math.ethz.ch" <bioconductor at="" stat.math.ethz.ch=""> > Subject: [BioC] edgeR - technical and biological replication > > Hi, > I have a RNA-Seq dataset produced on a Illumina GA and wonder how to > account for both technical and biological replication using edgeR. The > dataset consist of 48 individual samples (time series with 4-6 > biological replica per time point, 9 time points in total) which have > been sequenced in three GA runs (3 technical replica) using 12-plex > sequencing (4 lanes/run) - that is a total of 144 profiles representing > 3 technical replica and 4-6 biological replica. Simply adding the counts > from the 3 tech rep forming 48 profiles would be a start point, but it > would be more powerful to include the tech rep in the analysis as > well... > > Best regards, > Jakob > > > ############################### > R version 2.11.0 (2010-04-22) > x86_64-pc-mingw32 > > locale: > [1] LC_COLLATE=Danish_Denmark.1252 LC_CTYPE=Danish_Denmark.1252 LC_MONETARY=Danish_Denmark.1252 LC_NUMERIC=C > [5] LC_TIME=Danish_Denmark.1252 > > attached base packages: > [1] stats graphics grDevices utils datasets methods base > > other attached packages: > [1] edgeR_1.6.0 > > loaded via a namespace (and not attached): > [1] limma_3.4.0 tools_2.11.0 ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}