questions on amplicon-processing vs shotgun processing

[WSL]

Hello. I'm sorry that my question maybe a dumb question. I read an article from http://pathogenomics.bham.ac.uk/blog...on-processing/ and finding it so interesting! I would like to go back to my old runs to reanalyse the data using the shot-gun processing. But in order to do this, do I need my PTP to reprocess? Does anybody know how to set up the reprocessing with the shot gun processing?

Many thanks!

[ajthomas]

It's really not hard at all. The simplest way is to open either the D...imageProcessingOnly... dataset or the R directory in gs Run Browser. On the overview tab, there is a section on the right hand side with the heading "Run Processor Manager". Just type a name in the box, choose the processing type below, and hit the start button.

Alternatively, you can use the command line by typing "runAnalysisPipe" and then the R or D directory you want to process.

In my experience, I have found that even though I get more reads by using the shotgun pipeline, those extra reads aren't helpful in the data analysis. In fact, I found that there was some difference in homopolymer regions of some reads depending on which pipeline was used. For example, a particular sequence may have a homopolymer of 5 G's. When using the Amplicon pipeline, I might get 50 reads of that sequence, all of which have 5 G's, whereas with the shotgun pipeline I might get 70 reads, but only 35 have 5 G's and the rest have 4 or 6. Yes, that's right, some of the reads that were right with the amplicon pipeline were wrong with the shotgun pipeline. Of course, that was a while ago when I observed that, and I haven't confirmed that this still happens with any updates to the software, but that is what I saw back then. In short, I found that the extra data obtained with the shotgun pipeline didn't help at all, and in some cases made things worse. Maybe it will work better for you.

[WSL]

Thanks ajthomas! That was really helpful! I managed to reprocess the data with shot-gun processing and the total reads (passed filter wells) have doubled! However, I have now got another questions of my run http://seqanswers.com/forums/showthr...5418#post85418

[RCJK]

Hi WSL, depending on the type of amplicon and purpose of the project, you may also want to see how the data looks (at least the # of reads) using a modified amplicon signal processing. You'll get more reads than the default amplicon processing, but the read lengths may vary more. The way to do this is described in Roche's amplicon design guidelines.

[WSL]

Thanks RCJK! The modified amplicon signal processing sounds very appealing to me. I will reprocess my data and then compare it. Do you know what are the differences on the threshold of each filters (I assume there are major differences between different signal processing pipelines)?