Ion Torrent adapter heterodimer

[HMorrison]

We've been struggling to get fusion primers to work on IT system for 16s tag sequencing. Primary reason for using the system--don't do much genomics. Don't know why it took so long to check this, but it looks like the problem is not the fusion primers but the IT sequences themselves (see attached). So these work fine in adapter ligation, but not in PCR. We get faint, broad double bands, generally one at expected size and other higher. Not enough material to size select or even simply clean up.

Anyone else trying this approach? Clearly the 5' ends are going to have to be kept for binding to spheres, but maybe we can change the sequencing primer complementary part--?

Ion Torrent folks, please chime in...will also post to Ion Community.

[HMorrison]

So I take it from the 126 views and 0 responses that no one is using a fusion primer approach. Hmm. That's informative by itself.

[krobison]

There are a lot of folks using fusion primers; year end isn't a time to get fast responses!

I'm trying to understand where in the process you think you are generating these products -- clearly the IT sequences (which no, I don't love either) work in the normal protocol & in your fusion primers these shouldn't really matter anyway as these aren't 3' ends (plus, it would be surprising if they really dimerized productively given that mismatched 3' end -- unless you are using a polymerase with 3'->5' exonuclease activity)

How much have you tried to optimize the PCR conditions (temps, polymerases)?

It might help to post your entire fusion primer sequences rather than just the IT adapters. You might also consider sanger sequencing one of the unexpected bands to see what it really is.

[HMorrison]

There are a lot of folks using fusion primers; year end isn't a time to get fast responses!

Well, there were certainly a lot of *views* of the thread...

I'm trying to understand where in the process you think you are generating these products -- clearly the IT sequences (which no, I don't love either) work in the normal protocol & in your fusion primers these shouldn't really matter anyway as these aren't 3' ends (plus, it would be surprising if they really dimerized productively given that mismatched 3' end -- unless you are using a polymerase with 3'->5' exonuclease activity). How much have you tried to optimize the PCR conditions (temps, polymerases)?

The complication is that if we want to be able to compare across datasets generated on different platforms, we'd hoped to keep the PCR constant. It clearly will not be possible. The adapter ligation is slow and inefficient relative to using fusion primers.

Shauna suggested reducing cycles and extension time and we will try that next. Also gradient of annealing temps, but again we would like to be able to compare to 1000+ datasets we already have using 454.

It might help to post your entire fusion primer sequences rather than just the IT adapters. You might also consider sanger sequencing one of the unexpected bands to see what it really is.

Been there, done that...inconclusive. Have to clone and sequence, actually. We get products that have a bit of the target, not simply primer seqs. The specific part of the primer adds to the heterodimer match, certainly. There are no good alternatives if we want to target this region in all bacteria.

I am going to start looking at extending it based on a 200 nt read. Neither Illumina nor Ion Torrent are as readily adaptable to tag sequencing with fusion primers as we'd hoped.

[krobison]

I definitely understand the desire to not use ligation; it gets expensive & work intensive for large sets. And yes, I was one of the prior lurkers -- reading is faster than writing!

If you are using the same targeting region & PCR conditions as with your 454 primers, I agree it is curious you get these bands with one technology but not the other. Plus, inclusion of some payload would seem to rule out simple mischief with the fusion primers. You may be having processivity issues; consider using (if you aren't already) a polymerase such as Phusion which has a clamp engineered on. Trimming back extension times can help with non-specific priming, as can playing with annealing temperature. I'll confess a bit of confusion with not wanting to tinker with conditions; quality sequences are what I would see as important. In any case, you really need to rerun a good set of your prior samples on the new platform, just to understand the platform-specific error modes. I would think switching platform will have a far greater impact on trying to compare results than tuning amplification conditions.

One other thing to explore is oligo quality -- try once with PAGE purified oligos. Yes, they are painfully expensive but it might help you troubleshoot.

I can understand your issue with read lengths. You might also explore CCS sequencing on a PacBio instrument, which might get you the lengths you need. But, those are rare beasts (there are a few commercial providers).

[ECO]

This is news to me, but I didn't realize that Ion is (was?) using a combination of 454 and SOLiD adapters...

Googling the first and second sequences that you posted shows that the first is a 454 titanium adapter, and the second is the inner portion of the SOLiD P1 adapter

...probably needless to say these weren't optimized to work together in PCR, you will undoubtedly need to optimize your system to work with these primers.

Some other thoughts on the fusion approach...don't forget that you will be generating reads unidirectionally across the amplicon, which isn't a good thing given the context specific error of all these platforms. Ligation gets around that (producing symmetric reads), as does designing two primer sets per amplicon, but then you're up to 4 primers and two PCRs per amplicon.

Another reason (IMO) to avoid the fusion primer approach is that your thermocyclers are now churning out immense quantities of molecules that are perfect contaminants for your emulsions/flowcells. With the ligation approach your contamination worries, while not eliminated, are much reduced (as the only flow-cell/emulsion competent molecules are generated via ligation).

Lastly, staying with ligation lets you stay platform agnostic...and sequence the same amplicons on multiple systems. Not sure how important that is for you going forward...

Using a one-tube ER/AT/ligation approach (a la 454's Rapid prep), one can make the ligation step take about an hour with a single purification...

[HMorrison]

Problem resolved

The solution suggested by Shauna from Ion Torrent (on the Ion community site) worked. We cut down number of cycles from 30 to 22 and it eliminated the unwanted products.

And thanks for the other ideas, but we do want to sequence unidirectionally at least until we can cover the amplicon with paired end reads. And we have already thoroughly contaminated our lab with fusion-primer amplicons using 454 for microbial diversity studies .