Dear Suryavadhan for normalisation between samples, please use the method described in the DESeq vignette, rather than the (information-losing) method described below. For the non-unique reads, DESeq has no provision for fuzzy or fractional alignments. You'll have to make a choice, and provide actual counts. Hope this helps Wolfgang Sep/15/11 6:58 PM, Kayilai, Suryavadhan (MU-Student) scripsit:: > FOr the 6 sequenced samples, we ran alignments to get expression > estimates. The protocol is to align the reads, then count the number > of reads falling within the boundaries of the annotated genes, then > normalize with respect to the number of reads aligning in each > sample(not the sample length). The analysis also attempts to capture > the non uniquely aligning reads by estimating the unique read counts > for each gene, then apportioning the ambiguously aligning reads among > the potential sources based on the ratios of read counts among those > sources established by the less ambiguous readsie the first round of > apportioning assigns 2-mapped reads based on the unique alignments, > then 3-mapped reads are apportioned based on the adjusted read > counts, and so on). So, in the attached you'll see three sets of > columns for the samples, with those head "unique" giving the > per-million-reads-aligned normalized values for each samples uniquely > aligned reads, "apportioned" using the adjusted values as described > above, and "total" giving the number of reads aligned to the gene > models without regard to their uniqueness. Note that in all cases, we > consider only reads mapping to no more than 5 locations. Hence, the > values that are in non integer forms. Kindly help me through this > > Suryavadhan ________________________________________ From: Steve > Lianoglou [mailinglist.honeypot at gmail.com] Sent: Thursday, September > 15, 2011 9:42 AM To: Kayilai, Suryavadhan (MU-Student) Cc: > bioconductor at r-project.org Subject: Re: [BioC] Analysing RNA-Seq data > using DESeq package > > Hi Suryavadhan, > > On Tue, Sep 13, 2011 at 12:41 PM, Kayilai, Suryavadhan (MU-Student) > <skhx5 at="" mail.missouri.edu=""> wrote: >> I downloaded the DESeq package for the RNA seq analysis of the >> Soybean genes. The package is really helpful and easy to use. >> Thanks! I have a small doubt and it would be kind of you, if could >> help me figure out the same. The package works fine for the gene >> data with whole number or integer values. How can I run the >> analysis for decimal data as the class newCountDataset does not >> allow me to input decimal data. It would be great if you could help >> me through this. > > It doesn't let you put in non-integer data, because the models DESeq > uses to test for significance assumes count data -- as in, the > number of reads that align to a given region, which can only ever be > integers. > > What types of data are you trying to put in that are decimal values, > anyway? What does it represent? > > -steve > > -- Steve Lianoglou Graduate Student: Computational Systems Biology | > Memorial Sloan-Kettering Cancer Center | Weill Medical College of > Cornell University Contact Info: > http://cbio.mskcc.org/~lianos/contact > > > > > _______________________________________________ Bioconductor mailing > list Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor Search the > archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor -- Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber

Dear Suryavadhan for normalisation between samples, please use the method described in the DESeq vignette, rather than the (information-losing) method described below. For the non-unique reads, DESeq has no provision for fuzzy or fractional alignments. You'll have to make a choice, and provide actual counts. Hope this helps Wolfgang Sep/15/11 6:58 PM, Kayilai, Suryavadhan (MU-Student) scripsit:: > FOr the 6 sequenced samples, we ran alignments to get expression > estimates. The protocol is to align the reads, then count the number > of reads falling within the boundaries of the annotated genes, then > normalize with respect to the number of reads aligning in each > sample(not the sample length). The analysis also attempts to capture > the non uniquely aligning reads by estimating the unique read counts > for each gene, then apportioning the ambiguously aligning reads among > the potential sources based on the ratios of read counts among those > sources established by the less ambiguous readsie the first round of > apportioning assigns 2-mapped reads based on the unique alignments, > then 3-mapped reads are apportioned based on the adjusted read > counts, and so on). So, in the attached you'll see three sets of > columns for the samples, with those head "unique" giving the > per-million-reads-aligned normalized values for each samples uniquely > aligned reads, "apportioned" using the adjusted values as described > above, and "total" giving the number of reads aligned to the gene > models without regard to their uniqueness. Note that in all cases, we > consider only reads mapping to no more than 5 locations. Hence, the > values that are in non integer forms. Kindly help me through this > > Suryavadhan ________________________________________ From: Steve > Lianoglou [mailinglist.honeypot at gmail.com] Sent: Thursday, September > 15, 2011 9:42 AM To: Kayilai, Suryavadhan (MU-Student) Cc: > bioconductor at r-project.org Subject: Re: [BioC] Analysing RNA-Seq data > using DESeq package > > Hi Suryavadhan, > > On Tue, Sep 13, 2011 at 12:41 PM, Kayilai, Suryavadhan (MU-Student) > <skhx5 at="" mail.missouri.edu=""> wrote: >> I downloaded the DESeq package for the RNA seq analysis of the >> Soybean genes. The package is really helpful and easy to use. >> Thanks! I have a small doubt and it would be kind of you, if could >> help me figure out the same. The package works fine for the gene >> data with whole number or integer values. How can I run the >> analysis for decimal data as the class newCountDataset does not >> allow me to input decimal data. It would be great if you could help >> me through this. > > It doesn't let you put in non-integer data, because the models DESeq > uses to test for significance assumes count data -- as in, the > number of reads that align to a given region, which can only ever be > integers. > > What types of data are you trying to put in that are decimal values, > anyway? What does it represent? > > -steve > > -- Steve Lianoglou Graduate Student: Computational Systems Biology | > Memorial Sloan-Kettering Cancer Center | Weill Medical College of > Cornell University Contact Info: > http://cbio.mskcc.org/~lianos/contact > > > > > _______________________________________________ Bioconductor mailing > list Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor Search the > archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor -- Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber