454 GS FLX+ long reads question

[eladSeq]

Hi,

I am a newcomer to deep sequencing and wanted to get some feedback on my project. My library consists of ~18,000 variants of an 860 bp gene on a plasmid. Because of its length I'm planning to use the 454 GS FLX+ system (which should have reads up to 1000 bp). I was also informed that I must use the one-way sequencing design with the Lib-L chemistry for getting the long reads and 1 million reads on a full plate.

Therefore my plan is to prepare two amplicons using PCR with the Lib-L primers, one with the A primer on the 5' end, and one with the A primer on the 3' end, since I want reads from both ends. Does anyone know if these primers must be purified (HPLC of PAGE)?

Also, is it always necessary to purify the amplicons with the AMPure XP beads, or will a zymo column purification suffice? Especially if I don't see adapter dimers on an agarose gel or BioAnalyzer run.

Thank you,
Elad

[pmiguel]

You can probably get away with un-purified primers.

About purifying the amplicons: I would use every method at your disposal to select against primer dimers. Cutting them out of a gel and then double AMPure purifying them is not unreasonable. (You can do it after pooling to reduce the number of samples you have to work with.)

I think the problem is two-fold:

(1) There is a high fold difference in the amplicon size (800 bp in your case) vs the primer dimer size (~80 bp?). That is 100x. So even a barely visible primer dimer is a problem when considered on a molar level.

(2) Primer dimers can anneal back to full length amplicons after a denaturation, right? I mean, there is nothing to stop them doing so. Even after a size selection you will still have some glomming on to what you purify.


So, those 2 factors combined work against a 454 amplicon experiment because the primer dimers are small enough to hide unnoticed among your products. If you get enough of them, they cause problems given that they will PCR amplify so much more easily than your intended amplicons.

This effect would be exacerbated with FLX+ runs, I would expect.

--
Phillip

[MissDNA]

Is FLX+ already validated for amplicon experiments? Last I heard, it was not.

[pmiguel]

Is FLX+ already validated for amplicon experiments? Last I heard, it was not.

Good point. I don't know. (Our upgrade did not happen yet.)

Elad,
You might be better off just doing 2 shorter amplicons with FLX chemistry.

Anybody using FLX+ for amplicons?


--
Phillip

[MissDNA]

Ours was suppposed to happen this month but was postponed to January. I talked about amplicons a couple of months ago with a Roche representative and he said FLX+ was still not recommended for amplicons, and since no new bulletin or manual was released I am assuming it is not.

[ajthomas]

I haven't heard anything different concerning amplicons and FLX+ chemistry either. Saying that, however, I'm going to give it a try some time in the next few weeks. Not an entire run, mind you, just a couple of sections of a plate on which I'm running some other stuff. I'll try to remember to share my findings here.

[RCJK]

XL+ is still not recommended for amplicons. Another lab that has run them has said that they have had consistently worse results than when the same amplicons were run with the XLR70. Roche has lowered the pricing of my XLR70 kits to be the same as the XL+ kits so that I'll use them for amplicons. That said, I did just do 1/8th of amplicons on an XL+ run and am waiting for feedback from the customer. Also, I'm not sure if you need to do anything different for amplicon signal processing with the XL+ kits. With shotgun processing I had ~85% pass filter, then when I did amplicon processing, it went down to ~12%. With modifying the filters to trim from the 3' end instead of discarding the reads, it went up to ~18% pass filter. Also, the controls looked weird and gave really long read lengths (1-2KB) after amplicon processing, so I'm not sure what that's all about.

[ajthomas]

I've heard that from a couple of sources. In my case, my libraries consist of 6 different amplicons with very high polymorphism. That means that the strong synchronous signals normally seen with amplicons that make sequencing them more difficult for the GS FLX should be reduced compared to a lot of other amplicon projects. I'm hopeful that means it might work a little better for me. I hope it does because two of my amplicons are adjacent exons in a gene that I would really rather sequence together rather than separately. The two exons with the intervening intron is short enough that I'm only about 50bp short of sequencing it reliably with the FLX. I'm hopeful that I might be able to get those additional 50bp by using the FLX+. If not, oh well, I guess I'll have to wait for improvements to the system/reagents to allow it to work.

[madseq]

In principle, it is not possible to perform a XL+ run (400 cycles) with amplicons because there is no an Amplicon pipeline for 400 cycle runs. It is not possible to analyze an amplicon library properly after the run.

However...
For amplicon and shotgun combination in the same plate I think that it could somehow be possible to modify the .cwf files by deleting the images corresponding to the last 200 cycles and then try the amplicon pipeline. It is not very good idea to use shotgun analysis for amplicons. The Amplicon pipeline takes into account the fact that the sample will be of low diversitity, which means that a lot of wells will have the same signal at the same time.

Below you´ll find what I was told from customer support when I asked about this issue:


- FLX+ chemistry is not compatible with Amplicon sequencing
- FLX+ instruments are backwards compatible to Titanium chemistry, so you can still do Amplicon sequencing using a Titanium kit. Limitation is the pipeline, as SW 2.6 does not contain an Amplicon pipeline for 400 cycle runs
- FLX+ sequencing kit and Titanium sequencing kit is different- FLX+ runscript is 400 cycles, Titanium runscript is 200 cycles. You cannot start a 200 cycle run using a FLX+ kit
- Titanium (shotgun and paired end, NOT amplicon) can be run using a FLX+ kit without a benefit in readlength

[MissDNA]

About PE, I don´t see the benefit of running it with a FLX+ since the library prep protocol was not modified, so the libraries have still shorter fragments than the desired for a good FLX+ run. I really wonder if Roche is working on optimizing PE protocols since they are already obsolete and time consuming.

[pmiguel]

XL+ is still not recommended for amplicons. Another lab that has run them has said that they have had consistently worse results than when the same amplicons were run with the XLR70. Roche has lowered the pricing of my XLR70 kits to be the same as the XL+ kits so that I'll use them for amplicons. That said, I did just do 1/8th of amplicons on an XL+ run and am waiting for feedback from the customer. Also, I'm not sure if you need to do anything different for amplicon signal processing with the XL+ kits. With shotgun processing I had ~85% pass filter, then when I did amplicon processing, it went down to ~12%. With modifying the filters to trim from the 3' end instead of discarding the reads, it went up to ~18% pass filter. Also, the controls looked weird and gave really long read lengths (1-2KB) after amplicon processing, so I'm not sure what that's all about.

See also this thread:

http://seqanswers.com/forums/showthread.php?t=15838

Seems to imply that even using XLR70 kits, there may be issues with the really long read lengths if the instrument is FLX+.

Caveat being that Indiana may have modified their processing pipeline from Roche standards. I don't know.

--
Phillip

[pmiguel]

About PE, I don´t see the benefit of running it with a FLX+ since the library prep protocol was not modified, so the libraries have still shorter fragments than the desired for a good FLX+ run. I really wonder if Roche is working on optimizing PE protocols since they are already obsolete and time consuming.

You could tweak the shearing parameters for the PE protocol to favor longer inserts.

I would not hold my breathe for FLX+ protocols. I think much of the research effort given to the GS-FLX shifted to the FLX Jr as Illumina sucked up more and more of the next gen market place. This is what I think caused the 1-2 year delay in the release of FLX+ chemistry. Big mistake, in my opinion, because MiSeqs and IonTorrents will probably completely crush any burgeoning opportunities for Roche to gain a strong foothold in the desktop nextgen sequencer market. Whereas having FLX+ in place 1-2 years earlier might well have solidified the 454 niche in core labs.

If Roche Applied Sciences can hold on for a while longer, they may begin to see some pick up in sales as the heavily "early adopter" phase of next gen sequencing matures into something more mainstream. That process would be really helped by run processing software that did reasonable analysis of failure that could be used by cores to rapidly identify where issues arose. For example, currently you have to upload a run to tech support to get a global assessment of signal droop across a run. As a result Roche tech support has to waste a larger percentage of their time trying to troubleshoot issues that would be trivial for a core to deal with directly.

--
Phillip

[MissDNA]

We just found out our upgrade won´t happen till January, also there was some software problem with some upgrade that has been made around here. Something regarding to the equipment temperature, which made me think about some post I read here mentioning temperature problems. At this point, we are a bit scared of the upgrade since our machine and sotware are running really well as is.

We are purchasing an Ion Torrent, that hopefully will be installed begining of next year. We decided in favor of having an Ion before we knew about MiSeq. Even though I never worked with Illumina an know very little about their technology, I really think it could have been a better purchase. We will see.

[RCJK]

Some feedback from 454 re: amplicon processing after sequencing with XL+ kits:

Amplicon processing for this dataset can be done as runAnalysisPipeAmplicons like every other dataset. However, since long Amplicons are not supported, the software (as of v2.6) as it currently stands is not tuned to expect long Amplicons. Consequently, the pipeline along with parameters, thresholds, error checkpoints and options are not modified to reflect this and correct results may not be obtained. So, while it is possible to run the pipeline and get reads results may not all be of good quality. Please keep in mind that the results and / or subsequent comparisons may not be accurate.


What I am unsure of after this response is whether or not "short" or <600bp amplicons sequenced with the XL+ kit and processed with amplicon signal processing as previously done would/should be ok/accurate. I'm awaiting clarification on this, but not holding my breath for a straight answer.
The amplicons I mentioned a few posts above were short (~230bp-480, but mostly <300bp) and were emPCRd using the short amplicon protocols.
Otherwise, the shotgun libraries have been looking really good with modal read lengths of ~700-800bp. So apart from the first failed run, I'd say the rest have been successful.

[tokikake]

Something regarding to the equipment temperature, which made me think about some post I read here mentioning temperature problems. At this point, we are a bit scared of the upgrade since our machine and sotware are running really well as is.

That is very interesting! I wrote some weeks ago about our problems with the upgrade and last week we had a visit from the service and they replaced the PTP heater because of some temp problems...I heard about the same problems from another sequencing lab. Honestly, after all the troubles in the last weeks I would definitely wait with the upgrade.