Deadly Library

[GW_OK]

I've come across a bit of a mystery in my Core and I was wondering if anyone might have an idea as to what's going on.

I have a customer who prepared an exosomal RNA library using a kit from System Biosciences (SBI), specifically their XRNA Exosome RNA-Seq Library Kit. There was much consultation and whatnot but in the end they generated what appeared to be a proper library.

Upon submitting their library to my Core we did our standard Tapestation check and Kapa qPCR. Tapestation gives a single peak at ~152bp, right where the kit says it should be. qPCR says the library concentration is 2nM, exactly what I asked the submitting lab to normalize to (they quantified with Qubit).

We load the library on our Miseq for a v3 PE75 run with 20% phiX (an amount requested by the customer). The Miseq errors out on the focus step, no cluster detected. Having a look at the focus images shows they are completely black. In full disclosure this kit uses a special primer for read 2, but that is spiked into the read 2 primer in the Miseq cartridge at the appropriate concentration and shouldn't be a factor in clustering.

We call up TS and tell them what's going on and we run through all of the system checks we can run without an FS rep coming in. Everything checks out. Must have been a fluke right? Maybe bad reagent kit? Whatever's happening, the customer is in a hurry and needs the data. So....

We load up the library again, this time bumping phiX up to 30% as suggested by TS. Again, run errors out on focus with no clusters detected. All images black again as well.

Something's obviously going terribly wrong somewhere. Re-qPCR, re-Tapestation, still looks the same. Qubit quantification agrees with qPCR. As a test of the machine we load up the next person in the queue. It clusters beautifully and runs to completion without issue. Granted this was a PE250 v2 kit so we can't say it's a complete apples-to-apples comparison reagent-wise but I've never seen two kits fail back-to-back.

In my years of Illumina sequencing I've seen libraries that won't cluster and you only see phiX, but I've never come across a library that will not cluster AND will also keep phiX from clustering. Especially at 20 and 30% loading!

The customer contacted SBI and they claimed they only repackage reagents and never actually did any of the science to develop the kit. (Uh, what?)

Thoughts or suggestions would be appreciated.

[dfhdfh]

What is the solvent of the customer library?

Edit: Buffer may be the better word, as the solvent most likely is water

[GW_OK]

EB buffer. 10mM Tris.

[NextGenSeq]

I had the same problem with a library made by a commercial company which they gave us to sequence (even though they had their own MiSeq). The only way I was able to get ANY data was to spike-in 50% PhiX. I got a couple of million of usable sequence but the stats were terrible.

[bilyl]

I've come across a bit of a mystery in my Core and I was wondering if anyone might have an idea as to what's going on.

I have a customer who prepared an exosomal RNA library using a kit from System Biosciences (SBI), specifically their XRNA Exosome RNA-Seq Library Kit. There was much consultation and whatnot but in the end they generated what appeared to be a proper library.

Upon submitting their library to my Core we did our standard Tapestation check and Kapa qPCR. Tapestation gives a single peak at ~152bp, right where the kit says it should be. qPCR says the library concentration is 2nM, exactly what I asked the submitting lab to normalize to (they quantified with Qubit).

We load the library on our Miseq for a v3 PE75 run with 20% phiX (an amount requested by the customer). The Miseq errors out on the focus step, no cluster detected. Having a look at the focus images shows they are completely black. In full disclosure this kit uses a special primer for read 2, but that is spiked into the read 2 primer in the Miseq cartridge at the appropriate concentration and shouldn't be a factor in clustering.

We call up TS and tell them what's going on and we run through all of the system checks we can run without an FS rep coming in. Everything checks out. Must have been a fluke right? Maybe bad reagent kit? Whatever's happening, the customer is in a hurry and needs the data. So....

We load up the library again, this time bumping phiX up to 30% as suggested by TS. Again, run errors out on focus with no clusters detected. All images black again as well.

Something's obviously going terribly wrong somewhere. Re-qPCR, re-Tapestation, still looks the same. Qubit quantification agrees with qPCR. As a test of the machine we load up the next person in the queue. It clusters beautifully and runs to completion without issue. Granted this was a PE250 v2 kit so we can't say it's a complete apples-to-apples comparison reagent-wise but I've never seen two kits fail back-to-back.

In my years of Illumina sequencing I've seen libraries that won't cluster and you only see phiX, but I've never come across a library that will not cluster AND will also keep phiX from clustering. Especially at 20 and 30% loading!

The customer contacted SBI and they claimed they only repackage reagents and never actually did any of the science to develop the kit. (Uh, what?)

Thoughts or suggestions would be appreciated.

Have you thought about doing a bead cleanup of the prepared library before sequencing? Maybe something is carrying over.

Alternatively, perhaps you can mix in 50/50 the library that just worked, rather than using PhiX.

[nucacidhunter]

I assume that PhiX and input library is denatured separately and diluted with hyb buffer and then proportionally mixed. In this case it is less likely that any possible library contaminant would inhibit denaturation of PhiX due to dilution of contaminants. Once denatured library is loaded into flow cell it has to hybridize to flow cell oligos for bridge amplification and then hybridised with Read1 primer. At this step two events can result in sequencing failure:

1- Any oligo with full or partial complementarity to flow cell oligo or its adapter compliment especially if end modified can block or disrupt bridge amplification resulting in clustering failure.

2- The same goes with oligos with complimentary sequences to Read1 primer that can interrupt extension of primers during sequencing cycles.

An alternative to oligos is carry over of a blocked nucleotide from library prep step which would have similar effects. A clean up as suggested by bilyl is good starting point for trouble shooting.

Library prep with this kit follows a less efficient small RNAseq library prep with high number of PCR cycles in comparison with main stream kits. It also has a clean-up step after 3’ linker ligation which reduces the likelihood of 3’ linker carryover. I wonder if there is any trace of oligos in tape profile which can act as blocker during those two events.

[HESmith]

The other possibility (assuming that you add phiX to the sample before denaturation) is that the sample is highly buffered or acidic, which would neutralize the NaOH used for denaturation. Alternatively, the sample pH could be basic and, after denaturation, exceed the recommended limit and adversely affect hybridization to the flow cell (although that would probably require a fairly high pH to block hybridization completely). In either case, cleanup of the sample would solve the problem.

[GW_OK]

Thanks guys.

It's supposedly in EB buffer, so I doubt it's an acidic or high buffer situation. Also, wouldn't stray oligos adversely affect qPCR?

We do add phiX prior to denaturation. And I will be suggesting a bead cleanup to the customer once they return from vacation. There wasn't much to begin with and even less now after two loads so if anyone is going to be potentially losing the rest it will be them.

[bilyl]

The other possibility (assuming that you add phiX to the sample before denaturation) is that the sample is highly buffered or acidic, which would neutralize the NaOH used for denaturation. Alternatively, the sample pH could be basic and, after denaturation, exceed the recommended limit and adversely affect hybridization to the flow cell (although that would probably require a fairly high pH to block hybridization completely). In either case, cleanup of the sample would solve the problem.

Somewhat relevant: I have seen the occasional improperly made buffers. Things like 10mM Tris Hcl pH 8.5 suddenly became 100mM because someone was tired when dispensing from stock. If that's the case then it would neutralize most of the 0.1N NaOH.

[nucacidhunter]

The other possibility (assuming that you add phiX to the sample before denaturation) is that the sample is highly buffered or acidic, which would neutralize the NaOH used for denaturation. Alternatively, the sample pH could be basic and, after denaturation, exceed the recommended limit and adversely affect hybridization to the flow cell (although that would probably require a fairly high pH to block hybridization completely). In either case, cleanup of the sample would solve the problem.

If the library and PhiX were pooled prior to denaturation, HESmith’s explanation seems highly possible. An ssDNA assay after denaturation could be used to test this possibility.