dual indices?

[greigite]

Does anyone have experience with Illumina's new "dual indexing" method in the Nextera kits? I guess this is their new multiplexing strategy, with different indices on adapter 1 and adapter 2. Illumina assures me that these samples can be run in the same flow cell as "single index" samples but I would feel much better about this if I got some outside confirmation.

The following text is from the Nextera DNA sample prep guide:

The dual indexing strategy uses two 8 base indices, Index 1 (i7) adjacent to the P7
sequence, and Index 2 (i5) adjacent to the P5 sequence. Dual indexing is enabled by
adding a unique Index 1 (i7) and Index 2 (i5) to each sample from 12 different Index
1 (i7) adapters (N701–N712) and 8 different Index 2 (i5) adapters (N501–N508) for
the 96 sample Nextera Index Kit (FC-121–1012), and 6 different Index 1 (i7) adapters
(N701–N706) and 4 different Index 2 (i5) adapters (N501–N504) for the 24 sample
Nextera Index Kit (FC-121–1011). In the Index adapter name, the N refers to Nextera
sample preparation, 7 or 5 refers to Index 1 (i7) or Index 2 (i5), respectively, and 01–
12 refers to the Index number.

[koadman]

Bump!

I too would very much like to know how other's experiences have been with dual-index reads.

How were the read qualities?

[csquared]

They can be sequenced just fine in a normal single or paired end run, with or without the usual index read 1. The index read 2 is in the adaptor sequence at the opposite end of the fragment from the index read 1. If you don't need it, you won't know it is there and there should not be any issue with mixing dual index samples with single index samples in the flowcell provided the correct primer mix is used. Any samples that do not contain Index 2 will produce sequence off the adaptor during the index read 2, but that can just be ignored. The demultiplexing and analysis workflow is a bit more complicated, but not surprisingly so.

Index read 2 is a reasonably elegant solution that hasn't been particularly well implemented into the existing workflows on the HiSeq and GS2. MiSeq is better. The workflow changes are a little cumbersome and the current reagent kits don't have enough reagents to do a total of 223 cycles. Hopefully Illumina increases the fill on the 200 cycle kits rather than sticking with using 4 50-cycle kits like it is now.

IMHO, there are much easier ways to get to 96-indexes than dual indexing in its current form.

Since Illumina has done away with the extra base of sequencing during the dual index sequencing (it's only 8 cycles to read an 8-base index), I wonder if they will drop the extra base (currently 7 cycles to read a 6 base index) in the standard indexing run. Right now, if you don't do the extra base, the HCS software won't show the on-the-fly demultiplexing.

[pmiguel]

the HCS software won't show the on-the-fly demultiplexing.

on-the-fly demultiplexing? Are you saying this is currently available on the HiSeq? Or is it a MiSeq thing, only?

--
Phillip

[koadman]

Thanks for the really useful info csquared!

A further question: when sequencing mixes of single-indexed and dual-indexed libraries, does the index 2 read on the single-indexed samples still read into the adapter, e.g. yielding high quality base calls? Or do they just fail to prime and give low quality bases for those cycles?

[csquared]

Thanks for the really useful info csquared!

A further question: when sequencing mixes of single-indexed and dual-indexed libraries, does the index 2 read on the single-indexed samples still read into the adapter, e.g. yielding high quality base calls? Or do they just fail to prime and give low quality bases for those cycles?

I'm almost positive it reads the adaptor itself but would need to check the actual sequence. Since it is a monoplex, the read quality is not necessarily high. The elegant part of the paired-end version of the dual index is that it seems to use the P7 flowcell sequence as the primer to sequence index 2. If index 2 isn't there, it reads the adaptor sequence. The single-end kit contains an extra sequencing primer for index read 2.

Illumina has posted a lot of details of the process, but unfortunately, it is spread out. A lot is in the FAQ at the following link and is pretty helpful to appreciate how it works Illumina Dual Indexing

Some more information is in the updated HiSeq (or other platform) intrumet documentation, some in the cBot documentation and finally some in the Nextera documentation.

[csquared]

on-the-fly demultiplexing? Are you saying this is currently available on the HiSeq? Or is it a MiSeq thing, only?

Yup. On the HiSeq. It is part of the new SAV software (v1.8.4). For a run that has indexing, an extra tab (Indexing) is available in the SAV. It will show the counts per index in each lane during the run so you get an idea of the barcode distribution prior to the completion of the run. It might require the HCS upgrade too that allows dual indexing.

[pmiguel]

Yup. On the HiSeq. It is part of the new SAV software (v1.8.4). For a run that has indexing, an extra tab (Indexing) is available in the SAV. It will show the counts per index in each lane during the run so you get an idea of the barcode distribution prior to the completion of the run. It might require the HCS upgrade too that allows dual indexing.

Okay, thanks. I installed the new version of SAV. (Requires .NET Framework 4 to run. That is new.)

Pointing SAV at old runs does not give demultiplexing info. So I think you are right about the dual indexing HCS upgrade being required for it to work.

Just to be clear: the Indexing tab will show results for the old single indexes? (I.e, TruSeq indexes?)

--
Phillip

[csquared]

Just to be clear: the Indexing tab will show results for the old single indexes? (I.e, TruSeq indexes?)

It won't show results for old runs. Only runs that are configured as index runs using the new version of HCS to go along with the new SAV. If you are using the TruSeq indexes, you have use a 7-base index read for it to work under the default i7 indexing (I think it is i7, I always get the i7, i5 mixed up). Otherwise, you need to configure the indexing conditions using the 'Custom' option to match your conditions. It is a pretty useful feature, but it does have some quirks to it. For example, if you are using the TruSeq indexes and a 6-base index read rather than 7, the SAV won't pick up the indexes during the run unless it is configured via the 'Custom' option.

[pmiguel]

For example, if you are using the TruSeq indexes and a 6-base index read rather than 7, the SAV won't pick up the indexes during the run unless it is configured via the 'Custom' option.

By 6-base index, do you mean via a 7 base read (the last one for phasing-correction)?

I did notice that the crazy new additional indices in the v2 kits are nearly identical to what one would expect from the small RNA 48 index list -- except they have an additional variable base in the 7th position.

--
Phillip

[csquared]

By 6-base index, do you mean via a 7 base read (the last one for phasing-correction)?

I did notice that the crazy new additional indices in the v2 kits are nearly identical to what one would expect from the small RNA 48 index list -- except they have an additional variable base in the 7th position.

I should have said that when using the TruSeq indexes and you select the Single Index option in the new HCS, it will do 7 cycles of sequencing during the index read (6 to read the barcode and the 7th for phasing as you mentioned). If you explicitly set the index read to only 6 bases total (no phasing correction), SAV will not detect the barcodes.

Interestingly, if you do dual indexing, both index 1 and index 2 are 8 base reads to detect an 8-base barcode. No phasing correction in either read. We have not used an extra base for phasing correction in the index read for as long as I can remember. Just doesn't seem to make any difference.