Hi Syed, See the manual page for ?results and the contrast argument. The contrast argument is also discussed in the vignette. For the second question, it's best to fit over all genes and then subset. Mike On Apr 5, 2014 6:17 PM, "Mubarak hussain Syed" <mosvey@gmail.com> wrote: > Thanks Michael, > I have more questions regarding my Deseq2 analysis. If I have three > conditions, ( Transcriptome at three different stages during development), > Which > way is the best to get differentially expressed genes at three > developmental time points?. > And further if I am interested in transcription factors only: > 1. Is it better to take Transcription factor list counts only and perform > Deseq analysis on that list. > 2. or Use whole gene counts list and then filter Transcirption factors > from Significant results., which way is better and is statistically > appropriate?. > > Thanks, I appreciate your time and concern . > > Best regards > Syed > > On Mar 16, 2014, at 2:23 PM, Michael Love <michaelisaiahlove@gmail.com> > wrote: > > hi Syed, > > Yes your commands look correct. > > You say you want to narrow down the gene list. In my previous email I > suggested you could filter by effect size (and in DESeq2 v1.4 you can use > the LFC threshold argument to the results function). > > filtering on large, positive effect size would look like this: > > sig.large.up.results <- results[which(results$padj < 0.05 & > results$log2FoldChange > 1),] > > Mike > > > > On Sat, Mar 15, 2014 at 3:08 PM, Mubarak hussain Syed <mosvey@gmail.com>wrote: > >> Hi , >> I am using Deseq2 for differential gene expression calculations. I have >> two conditions Condition a (T) and Condition b (P), >> when I run the command using the following command lines, I get a long >> list of differentially expressed gene, more than 1000 unregulated and >> more than thousand down regulated. I am new to Des, do you think my >> command lines are good for my experiment and how could I >> narrow down my differential gene list. I appreciate your time. Thanks in >> advance, code is : >> >> > countsTable <- read.delim ("48_96_filtered.txt", header=TRUE, >> row.names=1) >> > pdata = data.frame(condition = factor(c( "T", "T", "T", "P", "P", "P", >> "P"))) >> > library (DESeq2) >> > dds <- DESeqDataSetFromMatrix(countData=countsTable, colData = pdata, >> design=~condition) >> colData(dds)$condition <- relevel(colData(dds)$condition, "T") >> >> > dds <- DESeq(dds) >> > results <- results(dds) >> > results <- as.data.frame(results) >> >> > sig.up.results <- results[which(results$padj < 0.05 & >> results$log2FoldChange > 0),] >> > sig.down.results <- results[which(results$padj < 0.05 & >> results$log2FoldChange < 0),] >> >> > sig.results <- results[which(results$padj < 0.05),] >> > sig.results <- sig.results[order(sig.results$log2FoldChange, >> decreasing=TRUE),] >> >> > plotMA(dds, pvalCutoff=0.05) >> >> > write.table (sig.results, file= "DESeq2_Sigresultsfiltered_48_96.txt", >> sep= "\t ", row.names=TRUE, col.names=TRUE) >> > write.table (results, file= "DESeq2_allresults_filtered_48_96.txt", >> sep= "\t ", col.names=TRUE) >> >> >> Best regards >> Syed >> HHMI-Institute of Neuroscience, >> 1254 University of Oregon, >> Eugene, OR 97403-1254 >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor@r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> > > > [[alternative HTML version deleted]]

Thanks Michael, I have more questions regarding my Deseq2 analysis. If I have three conditions, ( Transcriptome at three different stages during development), Which way is the best to get differentially expressed genes at three developmental time points?. And further if I am interested in transcription factors only: 1. Is it better to take Transcription factor list counts only and perform Deseq analysis on that list. 2. or Use whole gene counts list and then filter Transcirption factors from Significant results., which way is better and is statistically appropriate?. Thanks, I appreciate your time and concern . Best regards Syed On Mar 16, 2014, at 2:23 PM, Michael Love <michaelisaiahlove@gmail.com> wrote: > hi Syed, > > Yes your commands look correct. > > You say you want to narrow down the gene list. In my previous email I suggested you could filter by effect size (and in DESeq2 v1.4 you can use the LFC threshold argument to the results function). > > filtering on large, positive effect size would look like this: > > sig.large.up.results <- results[which(results$padj < 0.05 & results$log2FoldChange > 1),] > > Mike > > > > On Sat, Mar 15, 2014 at 3:08 PM, Mubarak hussain Syed <mosvey@gmail.com> wrote: > Hi , > I am using Deseq2 for differential gene expression calculations. I have two conditions Condition a (T) and Condition b (P), > when I run the command using the following command lines, I get a long list of differentially expressed gene, more than 1000 unregulated and > more than thousand down regulated. I am new to Des, do you think my command lines are good for my experiment and how could I > narrow down my differential gene list. I appreciate your time. Thanks in advance, code is : > > > countsTable <- read.delim ("48_96_filtered.txt", header=TRUE, row.names=1) > > pdata = data.frame(condition = factor(c( "T", "T", "T", "P", "P", "P", "P"))) > > library (DESeq2) > > dds <- DESeqDataSetFromMatrix(countData=countsTable, colData = pdata, design=~condition) > colData(dds)$condition <- relevel(colData(dds)$condition, "T") > > > dds <- DESeq(dds) > > results <- results(dds) > > results <- as.data.frame(results) > > > sig.up.results <- results[which(results$padj < 0.05 & results$log2FoldChange > 0),] > > sig.down.results <- results[which(results$padj < 0.05 & results$log2FoldChange < 0),] > > > sig.results <- results[which(results$padj < 0.05),] > > sig.results <- sig.results[order(sig.results$log2FoldChange, decreasing=TRUE),] > > > plotMA(dds, pvalCutoff=0.05) > > > write.table (sig.results, file= "DESeq2_Sigresultsfiltered_48_96.txt", sep= "\t ", row.names=TRUE, col.names=TRUE) > > write.table (results, file= "DESeq2_allresults_filtered_48_96.txt", sep= "\t ", col.names=TRUE) > > > Best regards > Syed > HHMI-Institute of Neuroscience, > 1254 University of Oregon, > Eugene, OR 97403-1254 > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]