Hi Goseq ers, I followed the user manual for Goseq and came up with the following plot, I am working with corn data, so I imported the lengths from Biomart and manually created the annotation. Can you please guide me if the Pwf plot is a good fit for my data, also is there a way to find which of gene-length and counts to use as bias.data. The plot is for gene length bias. I have 1802 DE genes and 28771 nonDE genes in my samples (edgeR output) . Also if the plotPWF function plots the pwf values, why only a handful of values can be seen in this plot? Please pardon my statistical ignorance. https://docs.google.com/open?id=0B_mk9qnFziV- SHE0LXQ0ZUFUb0dNMEEzN1UwVGhjQQ Many Thanks, Al [[alternative HTML version deleted]]

Hi Al, To me the plot looks like a perfectly good fit to the data. The thing is that it doesn't look like there is much gene length bias in the plot that you have shown. I'm not really sure why this would be as it is usually a strong effect but perhaps the biological signal is even stronger than this technical effect? You might just want to check the matching of your gene length annotation to DE genes, just in case. Cheers, Alicia > From: Alpesh Querer <alpeshq at="" gmail.com=""> > To: bioconductor at r-project.org > Subject: [BioC] Goseq plot > Message-ID: > <cao0xdqpci-dntu+yckafuf6h5wdjofj954ot+ktrto_ngu=xfa at="" mail.gmail.com=""> > Content-Type: text/plain > > Hi Goseq ers, > > I followed the user manual for Goseq and came up with the following plot, I > am working with corn data, so > I imported the lengths from Biomart and manually created the annotation. > Can you please guide me if the Pwf plot is a good fit for my data, > also is there a way to find which of gene-length and counts to use as > bias.data. The plot is for gene length bias. > I have 1802 DE genes and 28771 nonDE genes in my samples (edgeR output) . > Also if the plotPWF function plots > the pwf values, why only a handful of values can be seen in this plot? > Please pardon my statistical ignorance. > > https://docs.google.com/open?id=0B_mk9qnFziV- SHE0LXQ0ZUFUb0dNMEEzN1UwVGhjQQ > > Many Thanks, > Al > ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com

Hi Al, Even though the spline is fit on all the data (green line) the plot is made by binning the genes according the length or number of reads and calculating the proportion of DE in each bin. Therefore you get fewer points than the number of DE genes. As you know to just plot genes individually is just plotting 1 or 0 and doesn?t give you a feel for the proportions. Cheers, Alicia On 28/02/12 10:00 PM, "bioconductor-request at r-project.org" <bioconductor-request at="" r-project.org=""> wrote: > Message: 14 > Date: Mon, 27 Feb 2012 11:41:09 -0600 > From: Alpesh Querer <alpeshq at="" gmail.com=""> > To: bioconductor at r-project.org > Subject: [BioC] Goseq plot > Message-ID: > <cao0xdqnygpm=n5ilo4otcutfotyo4uyysg+agiv5v6dsgbgpmw at="" mail.gmail.com=""> > Content-Type: text/plain > > Hi Alicia, > > I checked the length annotation, it was fine. Also I looked at the pwf plot > for counts bias, it looks even flatter > (no bias). But I was wondering why there are only a few points on the plot, > if we looking at gene length vs proportion of DE genes? > As I mentioned in the previous email, I have 1802 DE genes and 28771 nonDE > genes in my samples. > > Here is the plot for the count bias. > > https://docs.google.com/open?id=0B_mk9qnFziV- aEhjeGZsY3BSVC1tMWUwTElzWENXQQ > > > Thanks, > Al ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com