How to find the "one" barcode?

[elgor]

Hi guys,

lets say we have 10 samples in a lane and only one of those samples has a barcode and I want to find the "one" sample which has this barcode, when I start the demultiplexing.

I know the HiSeq has a problem with less than four barcodes, but how can I force "configureBclToFastq.pl" to give me all samples which have an index sequence that looks more than nothing?

For some reason, we didn't really think about that beforehand.

Thanks in advance
André

[HESmith]

On the sample sheet for demultiplexing, enter only the one sample and barcode. All of the non-indexed reads will go to the Undetermined_indices directory.

[GenoMax]

To add to what HESmith said: you would not be able to separate the rest of the samples unless they have an internal barcode that you can post-process.

We had a discussion of the "index" and "barcode" terms recently in this thread. I am not sure which of the two applies in your case: http://seqanswers.com/forums/showthread.php?t=19214

[elgor]

Thanks for your answers.

Originally Posted by HESmith

On the sample sheet for demultiplexing, enter only the one sample and barcode. All of the non-indexed reads will go to the Undetermined_indices directory.

I did exactly that, but did get only a few reads with the searched TruSeq index sequence (and the undetermined reads of course). Not the expected 1/10 of all reads on this lane - rather 1/600. Is it technically possible at all to have only one index sequence read correctly by the hiseq 2000? Or will I get just Ns or wrong calls? I heard that two indexes are the minimum.

Originally Posted by GenoMax

To add to what HESmith said: you would not be able to separate the rest of the samples unless they have an internal barcode that you can post-process.

Right. the other 9 samples are indexed internally. I should be able to get them from the undetermined reads with a little script of mine. I did so before.

Originally Posted by GenoMax

We had a discussion of the "index" and "barcode" terms recently in this thread. I am not sure which of the two applies in your case: http://seqanswers.com/forums/showthread.php?t=19214

Interesting. Thanks for the hint. I'll read it!

[GenoMax]

Assuming that the point you mention below is not the cause have you thought about these two alternatives:

a. The quantitation of your indexed sample may have been inaccurate, which could lead to lower than expected yields.

b. Did you try allowing for one (or more) errors in the tag sequence in the de-multiplexing process to see if you are able to retrieve additional data. What do the "tags" in the "undetermined" pool look like?

Originally Posted by elgor

I did exactly that, but did get only a few reads with the searched TruSeq index sequence (and the undetermined reads of course). Not the expected 1/10 of all reads on this lane - rather 1/600. Is it technically possible at all to have only one index sequence read correctly by the hiseq 2000? Or will I get just Ns or wrong calls? I heard that two indexes are the minimum.