Hi Frederico- Looks like you've got it just about right — you want to use the bUseSummarizedOverlaps option. For paired end, you don't need to set the insert length (changed to fragmentSize moving forward) as with paired end data the size of each fragment is known. I will add the option to set the fragments parameter (in DBA$config$fragments), it should appear in the development version 1.9.9. Do let me know if you have any problems wit this, as this is a scenario we'd really like to see work and can help debugging if there are issues. Cheers- Rory From: Federico Marini > Date: Mon, 24 Feb 2014 14:51:49 +0100 To: Rory Stark > Subject: DiffBind and paired end data Dear Dr. Stark, I had the pleasure to attend ("back then in 2013") to the presentation you gave at the EBI Advanced Course for RNA-seq and ChIP-seq, where you introduced DiffBind and its usage examples. Now I moved from IMB - Mainz and I am working in the Biostatistics department of the University of Medical Center in Mainz as a research associate, where I also have the chance of doing my PhD. I contact you regarding indeed DiffBind, and the possibility of doing differential binding analysis for ChIP-seq paired end data. As one of our collaborators recently produced such datasets, and they would like to investigate the aspect of changes in the binding for TFs and histone modifications, I thought the DiffBind framework would be a very solid solution for analysis. The doubt I have, is whether DiffBind can use the information of the paired end data, and how. Ideally I guess the best way would be this: properly paired reads will be counted just once for each of the ends, and singletons(/not properly paired) will also count one. I was following the discussions around https://stat.ethz.ch/pipermail/bioconductor/2013-June/053394.html, and it seems that by incorporating the summarizeOverlaps functions it would be (almost-)possible, but I would like to double check it with you. bUseSummarizeOverlaps would be set to TRUE, DBA$config$singleEnd to false. Is there anything else I should take into account (e.g. the insert length in this case would be meaningful?)? Is there also a parameter to use the "fragment" parameter set to true in the summarizeOverlaps function? Thank you very much in advance for the attention, and thank you again for the nice package on top of it! Best regards, Federico -- Federico Marini, M.Sc. Medizinische Biometrie _____________________________________________ UNIVERSITÄTSMEDIZIN der Johannes Gutenberg-Universität Mainz Institut für Medizinische Biometrie, Epidemiologie und Informatik (IMBEI) Abteilung Medizinische Biometrie Postanschrift: 55101 Mainz Haus- und Lieferanschrift: Obere Zahlbacher Str. 69, 55131 Mainz www.imbei.uni-mainz.de Telefon +49 (0) 6131 17-7029 Telefax +49 (0) 6131 17-472433 E-Mail: federico.marini@unimedizin-mainz.de marinif@uni-mainz.de [[alternative HTML version deleted]]