Index read problem

[MLog]

Hi everyone, I'm a new user of Hiseq, this is a first time that I'm running a multiplex run and I have some concerns about Index read. Until the index read, the quality values were very high but after they dropped dramatically (see attached screenshot). Does it mean that I will not be able to demultiplex my samples?
The multiplexing was done for one lane only (adapters 2, 4, 5, 6 and 12) and for all other lanes only one sample per lane was used. Can this be an issue? As far as I can conclude from this topic: http://seqanswers.com/forums/showthr...quality&page=2 this can also affect the quality of Read 2?
For library preparation I used TruSeq DNA Sample Prep Kit, if this matters.

[pmiguel]

You are probably fine. For the single index lanes, the reads that fail to de-multiplex will end up in the "unassigned" folder -- you can recover them from there.

The focus carrying over to read 3 issue was solved with a firmware upgrade some months ago. Basically the instrument will always reset its focus at the beginning of read 3.

--
Phillip

[MLog]

Thanks, Phillip! Indeed, I checked it today and the Read3 seems to progress normally, with % of Q30 bases of about 95.

And what about the multiplexed lane? Is it possible to retrieve some data from it? I assume that if index read quality is poor (only about 40% of Q30 bases) then there will be a lot of errors in the index sequences. CASAVA allows either no mismatches or 1 mismatch. What if we run some custom demultiplexing script, setting acceptable mismatch number to 2? Will it help or it will rather mix up the reads from different samples?

[HESmith]

Check the %Q30 for the lane that you multiplexed, rather than all lanes (as shown in attached figure). I suspect that it's fine.

[westerman]

[As HESmith implied in a post made while I was composing mine] Maybe you can send up a snapshot of the multiplex lane only? The snapshot you sent was all lanes thus is hard to tell what the quality of the index is like in the multiplex lane.

Your barcodes are a good mix. None of them are very close to each other; i.e., have similar bases. On the other hand with a mismatch of 2 you can get into problems with choosing which barcode to choose. For example a read that has 'CGATCT' could, with 2 mismatches, go to either barcode #2 or #4.

A better bet, if possible, is to just not use all 6 bases. It is possible in CASAVA to mask out which bases to use in the barcode. I've done this when the quality in some of the index bases was poor but the quality was high in the other bases. It is possible to even go down to one base for demulitplexing if said base is very high quality -- obviously with no mismatches allowed. Desperate times require desperate measures!

[pmiguel]

By the way, you can right click on most SAV graphs and the context menu will offer the choice "copy to clipboard". Might be easier than taking a picture of the screen...

--
Phillip