[WPAFB]

We got Roche to come out to service the instrument two weeks ago. But we had to call and nag and pester them for weeks to get them to respond. And the tech only had to drive 50 miles to get here. *sigh*

[kmcarr]

Is anyone out there actually using the amplicon processing pipeline but us? HMP sequencing centers' pyrotag 'amplicon' sequencing SOP (shotgun library kits and shotgun processing to get amplicon data) hurt us badly...

Hilary,

You are not alone. We ended up with a number of pissed off customers because it didn't work. That HMP 16S protocol was a debacle.

Maybe we should start a support group.

[HMorrison]

Hilary,

You are not alone. We ended up with a number of pissed off customers because it didn't work. That HMP 16S protocol was a debacle.

Maybe we should start a support group.

I am so ready to just give up on this company and turn my energy towards making the HiSeq work well for us...the PGM has an unfortunately high rate of non-random error for amplicon reads and most of the PGM team came from Roche anyway. We're certainly not going to bother with the FLX+ upgrade now. Would just add more variables.

[MissDNA]

We are still waiting on supplement CB+ to perform our 1st Plus run. I am super curious. It is amazing that we have been waiting almost 3 months from the day they delivered the first reagents till now to get a complete kit in the lab.

About amplicon 16S we only performed a few runs, they worked all right and our bioinformatician used the amplicon pipeline for analysis. He did looked at the data produced with SG pipeline and said it was not trustworthy.

[HMorrison]

We decided NOT to upgrade. 90% of what we do is amplicon work, and I don't believe there will ever be a FLX+ amplicon protocol or updated processing pipeline for it.

The picture shows the distribution of read lengths for the two pipelines. Amplicon reads SHOULD form a tight distribution at a long length. Shorter reads didn't make it as far with good quality--why would someone want them? Rhetorical question.

Can't believe you've been waiting 3 months for one component though. My sympathies.

[MissDNA]

We work mostly with SG and PE, although recently we have had several inquires about amplicon 16S. When people ask about that, I say we can do it using regular Titanium chemistry.

The peaks you showed are impressive. Since SG filters are less stringent than amplicon, we obviously get more reads, but as your graph shows well is mostly garbage. What amazes me is that Roche's tech support actually recommends the use of SG pipeline for 16S analysis.

Waiting for reagents has been a constant since we started operating in 2008. During all this time we hardly ever got a complete kit at once. Eventhough we have told them a million times that we can only run if all the components are here, they seem don't care. This is awful because sometimes when the missing item arrives, some other component of the kit expires. They do replace the expired reagent in this case, but the replacement is usually not immediate and that generates more delay and mess, because Roche Brasil does not usually have reagents in stock.

About the upgrade, since our machine was upgraded our XLR70 got better, not sure if is coincidence or not. Also, we started reprocessing some runs using SW v2.6 and we got more data. We still have to check if this increase of data is of good quality.

Our beads are ready to run but where is supplement CB+???

[jswebb2]

Our upgrade kit sat on the lab floor for 4 months--now machine and kit at factory being upgraded, and they have delayed its return date by a month. Our sales rep left in Oct--and no one from Roche 454 has called us since.

[MissDNA]

We finally were able to make our 1st Plus run today. It takes about 24 hrs to run an analyze, so only on Monday I will see the data. I am so curious!

Our machine was upgraded in January, but we were waiting for the upgrade to happen since last Oct.

[MissDNA]

Our 1st FLX+ run was a total disaster. I submited the reports to Roche this morning and they are analyzing it. Hopefully they will come up with an explanation.

The libraries seemed fine on the agilent. I even showed them to Roche's FAS before hand and he said they were ok for sequencing. Titration was fine and enrichment in the correct range. So I assume something happened during sequencing/processing...

[MissDNA]

So I already got the feedback on our failed first Plus run.
They put the blame on our library, saying that is full of small fragments. It is pretty much known to everyone now that amplicon libraries suffer from unseen short fragments. I was not aware that could happen in SG preps. Out Agilent traces showed libraries in the expected size range and with 0% fragments bellow 500 bp.
We have always have used Bionalyzer evaluate library quality, and that was by Roche´s recommendation. What to do now that we cannot trust completely in the peaks we get?

[RCJK]

can you post an image of the bioanalyzer trace?

[pmiguel]

So I already got the feedback on our failed first Plus run.
They put the blame on our library, saying that is full of small fragments. It is pretty much known to everyone now that amplicon libraries suffer from unseen short fragments. I was not aware that could happen in SG preps. Out Agilent traces showed libraries in the expected size range and with 0% fragments bellow 500 bp.
We have always have used Bionalyzer evaluate library quality, and that was by Roche´s recommendation. What to do now that we cannot trust completely in the peaks we get?

Are these PCR amplified libraries (eg, paired end)? If not then I don't see a mechanism for "unseen short fragments" to hide away. They might be generated during emPCR however (primer dimers).

If they are paired end libraries and were PCR amplified then you can see the "hidden" small fragments by denaturing an aliquot (dilute to >5 ul in water or high quality formamide, heat to 95 oC for 5 minutes, then quick chill on ice) and then loading it on an RNA pico chip. Ignore anything larger than 500 nt, but see if you see shorter, especially much shorter fragments there. Remember to account for the molar concentration. Even if you only see 10% of the fragments in a peak at ~100 nt, remember that there are 10x as many molecules there as would be in a similar area peak at 1000 nt.

How did your control beads look?

--
Phillip

[MissDNA]

They were RL. However the original samples were product of a WTA. I don´t know exactly which protocol they used to do it but the cDNA that was sent to us was amplified. So in a way, we probably have a lot of the same thing like in an amplicon sequencing. Since GS FLX+ is not meant to sequence amplicons, it can be not ideal to sequence our sample. I threw my hypothesis to Roche´s support and I am waiting on his feedback. Because I am definitely not convinced by the short fragment story.

Btw he gave me a protocol for a Procedure for the evaluation of amplification after emPCR. It is oficial but I had never heard of it. I have attached it because it can be usefull for someone else.

I have also attached the bionalyzer trace, so you guys can have a look.

[MissDNA]

Phillip, the controls looked smaller than they should be for a FLX+ run: CATG median 350 bp and ATGC median 571 bp. Our library median was 117 bp, and only 20,42% passed filter wells. Most reads failed on the dot, but a lot of short quality as well.

[pmiguel]

Roche makes it insanely hard to trouble shoot bad runs by burying most of the information needed so deep you need special software to uncover it. But a Roche FSE showed me that using gsRunBrowser on the "Wells" tab there is a drop down menu just below the "Well Categories" that you may never have changed from its default of "Status". There are some other, potentially useful, menu entries like "Carry Forward" and "Incomplete Extension".

Did you use the new GS-FLX+ amplification protocol for your emPCR? Mostly the same, but a very different thermal cycler program:

1x 4 minutes at 94°C
50x 30 seconds at 94°C, 10 minutes at 60°C
10°C on hold

Whereas the GS-FLX protocol called for:

1× 4 minutes at 94°C
50× 30 seconds at 94°C, 4.5 minutes at 58°C, 30 seconds at 68°C
10°C on hold

Anyone have an idea as to why on earth the recommended library molecule sizes are so high for the FLX+? Read lengths max out below 1 kb -- what is the point of having molecules 2x that length? Seems like the PCR yields would just be lower.

--
Phillip