Reanalyzing image files

[Benny Chain]

Can anyone give me any help on how to reanalazye MiSeq runs, using saved image files ? I can't find any help on this in the Illumina site or elsewhere. We run RTA 1.16.18 and have saved a run which I would like to reanalyze with different phasing parameters etc. Can one install RTA as a stand alone programme on another computer (i.e. not the one running the sequencer) and run it on a set of image data already stored ?
Many thanks for any help Benny

[simonandrews]

I didn't think you could save images on a MiSeq (at least not without hacking around in config files in an unsupported manner). The images which are saved by default are not the full raw data and you can't re-analyse from them.

If you have persuaded the machine to really save raw data then you should be able to run it through the old GA OLB pipeline but I suspect this isn't actually going to be an option.

[Benny Chain]

Thanks for the reply. They look like image files ! How do I tell whether they are the real thing ? They certainly occupy a lot of space : about 350GB on one run. Benny

[JamieHeather]

I should say, we tried to save raw images (tiffs) and cluster intensity files (cifs) saved from this run, but for some reason only the tiffs saved, despite having the SaveIntensities flag set to true in the run parameters file.

We're just having trouble re-running RTA to regenerate the tiffs (and subsequent bcls), and our nearest MiSeq's just been upgraded and doesn't have all the software back on it yet.

[GenoMax]

I should say, we tried to save raw images (tiffs) and cluster intensity files (cifs) saved from this run, but for some reason only the tiffs saved, despite having the SaveIntensities flag set to true in the run parameters file.

Do you manually delete files from a previous run before starting a new run?

We're just having trouble re-running RTA to regenerate the tiffs (and subsequent bcls)

Did you edit the "MiSeq.configuration.xml" file in the "RTA/config" folder and amend it with fixed matrix/phasing settings (e.g. from a good PhiX run)? I assume you are trying to re-run RTA manually after the edits?

Echoing Simon's sentiments I strongly recommend working with your local FAS and Illumina tech support since this would fall in the advanced "intervention" category and can potentially mess something up on the MiSeq.

[JamieHeather]

I should say, we didn't actually perform this run, it was done by our university's sequencing centre. So the rest of what follows is just to the best of my knowledge...

I don't believe they typically store image (or indeed cluster intensity) files - we only did this as a previous run trying out a new strategy had failed, so we wanted to keep all doors open.

That's right, we are planning to try and use the Loman/Quick trick and re-run the analysis with old PhiX settings. I haven't yet modified the MiSeq config file, but I plan to, once the MiSeq machine is up and running. I would prefer to do these on a separate machine, but there doesn't seem to be a MiSeq version of RTA available for download, and it's not bundled in with Reporter (i.e. there isn't a RTA/config folder), so it looks like doing it on the actual MiSeq machine might be the only option.

We've certainly been told by Illumina that running the data through GOAT won't work, but I'm hoping either re-running RTA with different settings, or running the cifs through an alternate basecaller (maybe AYB) might gain us something.

[GenoMax]

We've certainly been told by Illumina that running the data through GOAT won't work, but I'm hoping either re-running RTA with different settings, or running the cifs through an alternate basecaller (maybe AYB) might gain us something.

Can you tell us what the problem is with these runs?

Did you get no data, less data than expected, data with bad quality or a combination thereof? It sounds like the default iteration of RTA did complete on these runs.

[JamieHeather]

We've mentioned it elsewhere on here before (http://seqanswers.com/forums/showpos...0&postcount=20), but a quick summary...

We're trying to sequence variable amplicon libraries using a multiplex of 3 custom primers.

In our first run (just the libraries and 3 custom primers) we got back essentially no data; the vast majority of the clusters (of which there were very few, despite not looking terrible in the thumbnails) failed to pass filter, so there was very little data, most of which was of poor quality. There was however a few hundred to a few thousand reads off one of the sequencing primers in a few of the indices (certainly not all of those that should have been detected by it).

Our second run (where we saved the images/tried to save the cifs) had a 30% PhiX spike in, and we spiked our custom primers into the SP1. Here the data looked a lot better; more clusters, nicer intensities, higher quality. Very little of it got assigned to an index - it sadly turned out to pretty much all be PhiX, having as much or even less of the desired amplicon sequence than the first one.

Illumina seem to think it's just the primers, but the fact that in the first run we had some reads off one of them at least makes me think at best it's not the only problem...

(edit)... hence wanting to mine the images we have for any trace of the intended sequences, to see if we can narrow down what might have gone wrong/salvage anything that there is to be salvaged.

[GenoMax]

There are at least 2 other threads that were active in late October on this topic (you may have seen them).

We have been working through this as well. The directions we received from illumina tech support for using fixed matrix/phasing settings were different than the ones mentioned in Nick Loman blog article. [We have a v.2 MiSeq with the latest MCS installed].

We have not had much of a problem with cluster registration. When we run the same sample with the default XML the quality values drop after about 100 cycles (worse for read2, a 2 x 250 cycle run) but with a fixed-matrix XML file the quality drops can be stretched out to some extent.