The folks studying Ultra Conserved Elements (UCE) developed Splitaake for using self synthesized indicies and the deconvolution software.

https://github.com/faircloth-lab/splitaake/

I think the max # is several 1000s.

Unless Illumina has a new kit out, believe you will find that "384" kit you have is enough reagents for doing 4 sets of 96, not 384 unique combinations.

Unfamiliar with the two methods mentioned above, but the HiSeq and MiSeq are more sensitive to "randomness" in the first 5 base pairs of the read than the old GAII due to how they call clusters - that is to say, if you have a lack of randomness in those first five cycles followed by divergent sequence, those clusters will fail to pass filter. There is a very good explanation of this, as well as some barcode generation scripts, at: http://comailab.genomecenter.ucdavis.edu/index.php/Barcodes (Comal lab wiki)

Also, n.b., there is some proprietary modification to the Nextera indices that Illumina will not disclose. Unsure of what the effect or benefit is but have heard anecdotal evidence that synthesized copies of the standard Nextera indices do not peform quite as well as the "store bought" Illumina variety. Would be very interested if you find any method for reliably indexing >96 samples in a single MiSeq/HiSeq lane using Nextera or NexteraXT.

Best,

S. Wesley Long

You need to take advantage of dual indexing. Then more than 1000 indices is no problem.

Look here if you are interested in an amplicon strategy:

http://seqanswers.com/forums/showpost.php?p=94701&postcount=4

Regarding Wesley Longs answer:

1.The updated MiSeq is not as sensitive to low complexity samples anymore

2. If you look for more than 1000 indices, low complexity should not be an issue in the index sequences.

We use Access Array which is the most convenient and reliable system to prepare large libraries in a format 48 patients for 48 amplicons. Depending on the coverage needs different access array libraries can be pooled in a single MiSeq run by pooling same patients with more amplicons or more patients with the same amplicons.

It is correct that the updated MiSeq is not as sensitive to low complexity samples anymore.

Thanks for the advice everybody. The new Nextera Index kit does allow multiplexing of 384 samples. The most I would like to run would be 768 (8 96-well plates). I'm already adding a variable 0-3 N region between my target primer and the Nextera transposase sequence to deal with the library diversity issue. If I were to add barcode primers with either fewer or more bases than are present in the Nextera adaptors would this help with the diversity issue by shifting the reading frame on some of the samples?

Illumina's 384 sample kit provides eight i5 primers and twelve i7 primers. 8x12=96, the maximum number of samples you could pool in one run and subsequently demux. We have designed and tested many more than this and have obtained good data for 1400 samples in one run. The MiSeq could demux these with no problem. We are continuing to push the limit, which is partly a matter of finding software that can demux that many barcodes. We plan to publish the details so others can take advantage of this.

Hi, I have a slightly different question: I want to sequence about 1000 samples for dystrophin after gene capture. 

Is there any supplier that has 1000 barcodes so that I can run all of the samples in one lane, and if not, what is the highest number of barcodes I can assign to my samples?

Thanks

Iddo

You can make your own indexes. 

How?