Serial Illumina Sequencing (SI-seq) protocol for 16S - will it work on the MiSeq?

[stan1991]

Hi everyone,

I am looking to carry out a metagenomics project on the MiSeq or the HiSeq. I've done some reading around the subject and I found a paper by Maughan et al. that describes a protocol that uses 2 primer pairs introduced at different time points during the sequencing of the V567 region on the GA-IIx. When the second primer pair is introduced the first primer pair is removed by denaturation. The sequencing steps basically go like so:
1. Illumina primer used to sequence barcode - 8 cycles. Removed by denaturation.
2. Custom primers introduced. Forward primer anneals upstream of V5 and reverse primer anneals downstream of V7. After 36 cycles these are denatured.
3. New set of custom primers introduced. Forward binds upstream of V6 and reverse anneals downstream of V6. After 36 cycles they are denatured.
4. Fragment flips and is regenerated then the process is repeated one more time.

I was interested to know whether or not this protocol would work on the HiSeq or MiSeq?
I know there are reservoirs where custom primers can be added on the Illumina cartridge and I thought these could be used in addition to the other primer reservoirs to carry out this protocol. The sticking point for me is how the primers could be removed by denaturation?

Thanks for reading/replying.

Stan

[microgirl123]

Why don't you just make an amplicon with the regions that you are interested in (I know you can get V4V5 into about 400 bp) and then just sequence that? The scheme that you're proposing doesn't seem right to me. The MiSeq (and the HiSeq as far as I know) only use one primer at once - it's not like PCR! Have you seen the Caporaso 2011 paper for 16s metagenomics on the MiSeq? You can adapt that method for a longer insert size now that the MiSeq can accept longer reads.

[stan1991]

Thanks Microgirl.

I was looking to sequence over 3 variable regions (V123) in order to classify down to species level. The MiSeq will not cover this region - even with 2 x 250bp!! So I thought the SI-seq protocol could be a good option.

I have looked at the Caporaso protocol and thought about adapting it to sequence longer read lengths but due to the drop off in quality after 200bp we have decided it will not give enough sequence to classify to species.

Another idea we had was to PCR out a longer fragment from 16S (e.g. from V1 to V4) then sequence in from both ends to cover the V12 region with the forward read and the V4 region with the reverse read using the 2 x 250bp. Without any sequence overlap!! I think there would be problems with chimerism because we are trying to PCR out a longer fragment. I'd be interested to know what people think of this idea and of any issues you forsee with it??

[mcnelson.phd]

Hi Stan,

I've talked with my FAS about the whether or not you can do that protocol on the MiSeq and he said that while it is possible, you would have to execute a lot of custom recipe commands and Illumina would absolutely not support you in it.

We've done v4-5 and overlapped the reads just fine, I know some other people on here are trying v1-3 but I'm not sure how far they've progressed on it. We've seen that the v4 is generally sufficient for most projects unless you have a legitimate reason for needing to sequence more than one v-region.

[LVAndrews]

Illumina did announce that 2x300 kits will be available for the miseq this year with 2x400 kits by the end of the year. I know you shouldn't count on it until you see these kits, but if you can wait a few months, you might not have to change your protocol much. Perhaps for now you can target V4 like the rest of us and revise your data later on with the longer reads.

Alternatively, I have tried running our miseq for a longer read 1 rather than joining reads (I don't really see a difference in the data from joining reads across V4). Perhaps you could find a mode that will work well for you (using a 500cycle kit?), but given the q-score drop-off near the end of each read, prepare to give up about half your reads or more each run. One thing I have noticed is that community representation shifts substantially depending on the quality of reads you split out your libraries at. The other thing that seems to shift community representation in my experience is overamplifying your libraries (more than ~20 cycles, perhaps less).

[stan1991]

Having asked Illumina about SI-Seq and discussed it with my colleagues we've decided the protocol is nearly impossible on the MiSeq =(. Which is a shame because it seems like a cool protocol.

We've decided to do the 2x250bp reads instead for 16S sequencing!!

Thanks for your replies. They've helped loads =)