Extreme Density Gradients using HiSeq v3 chemistry

[nkondo]

We had 2 runs in Nov. Flow cell lot is no. 10112707 and reagent lot no. is 20006210 in both runs. Both runs showed "usual" density gradient.

[scotto]

In recent months are you seeing any v3 reagents that do not show the cluster density artifact?

The last 4 v3 flow cells we clustered show this effect. We're having to run 2 flow cells to get 1 flow cell worth of data...

--
Phillip

Since we first noticed this in mid-January, all of our paired end v3 runs (~35) have been affected to some extent, but the extent of the gradient varies from almost normal looking to horrible. The worst lanes have intra-lane densities from ~100 k - >1.2 million. Our v3 single end reads have been normal.

Another NGS group on campus is seeing this in their v3 paired end runs as well, although they're not running at the volume we are. The third NGS lab on campus is running v4 chemistry with no problems.

[pmiguel]

We have been seeing this for months. Tech support says it is v3 SBS reagents, but they can't narrow it to specific lot numbers.

In recent months are you seeing any v3 reagents that do not show the cluster density artifact?

The last 4 v3 flow cells we clustered show this effect. We're having to run 2 flow cells to get 1 flow cell worth of data...

--
Phillip

[pmiguel]

We have the same issue since Dec. last year. Our issues happened with flowcell lot number 2XXXXXX. Does anyone see this issue with flowcell lot number 1XXXXXXX?

I did see one 1XXXXXXXX flowcell with this cluster density gradient issue. However it was only present in 3 of the 8 lanes.

--
Phillip

[scotto]

We have been seeing this for months. Tech support says it is v3 SBS reagents, but they can't narrow it to specific lot numbers.

[nkondo]

We have the same issue since Dec. last year. Our issues happened with flowcell lot number 2XXXXXX. Does anyone see this issue with flowcell lot number 1XXXXXXX?

[pmiguel]

Sorry, I meant the actual sequencer itself had problems with the pumps and Peltier. The easiest way to test this would be to do a PhiX run on a preclustered and R1-primed flowcell that was generated at Illumina.

Thanks for the advice. Tech support seems certain that this is a chemistry issue. So, I think the problem is that even if they clustered the flowcell themselves, they might get a bad reagent kit and see a cluster gradient.

--
Phillip

[bilyl]

On the cBot? Yes, that is what we thought when we saw this last month. Engineers came out a couple of times. They also looked at the HiSeq.

Intensity looks even across tiles, but looks higher on the lower surface than the top.

Illumina tech support is apparently seeing a couple of these a day--always associated with v3 HiSeq chemistry.

No one else is seeing this?

--
Phillip

Sorry, I meant the actual sequencer itself had problems with the pumps and Peltier. The easiest way to test this would be to do a PhiX run on a preclustered and R1-primed flowcell that was generated at Illumina.

[pmiguel]

Phillip,

Could you clarify something please. The current Illumina HiSeq - High Output chemistry is v4. Is it v4 that you are running or older v3 kits (which were called TruSeq v3).

No, we can't run v4 -- the serial number on our instrument was too low. We are running v3.

--
Phillip

[kmcarr]

Illumina tech support is apparently seeing a couple of these a day--always associated with v3 HiSeq chemistry.
--
Phillip

Phillip,

Could you clarify something please. The current Illumina HiSeq - High Output chemistry is v4. Is it v4 that you are running or older v3 kits (which were called TruSeq v3).

[pmiguel]

How do the intensities scale on the flow cell? Has Illumina tried a PhiX run?

We had an issue similar to this and I believe it turned out to be a combination of flowrate and Peltier problems.

On the cBot? Yes, that is what we thought when we saw this last month. Engineers came out a couple of times. They also looked at the HiSeq.

Intensity looks even across tiles, but looks higher on the lower surface than the top.

Illumina tech support is apparently seeing a couple of these a day--always associated with v3 HiSeq chemistry.

No one else is seeing this?

--
Phillip

[bilyl]

We are seeing a high proportion of HiSeq v3 High Output chemistry runs display very low cluster densities at the top of flow cells. The first couple of rows of tiles at the bottom of the flowcells cluster at normal densities whereas the top 4-or 5 rows of tiles on the top of the flowcell -- especially the top surface are very low.

By "very low" I mean below 60 thousand cluster/mm^2 on the top surface and below 300 thousand cluster/mm^2 on the bottom surface.



Quality doesn't otherwise seem to be impacted.

As of a month or so ago, we thought this was an issue with our cBot. But as of last week, tech support appears to be attributing all cases of this issue to reagents. Illumina is sending us extra reagents to compensate for the loss of reads. That is, we are running two flowcells to collect one flowcell's-worth of data.

Since both of our last 2 runs show this phenomenon, others must be seeing it as well. Anyone figured out a work-around that allows the normal cluster density to be restored?

--
Phillip

How do the intensities scale on the flow cell? Has Illumina tried a PhiX run?

We had an issue similar to this and I believe it turned out to be a combination of flowrate and Peltier problems.

[pmiguel]

We are seeing a high proportion of HiSeq v3 High Output chemistry runs display very low cluster densities at the top of flow cells. The first couple of rows of tiles at the bottom of the flowcells cluster at normal densities whereas the top 4-or 5 rows of tiles on the top of the flowcell -- especially the top surface are very low.

By "very low" I mean below 60 thousand cluster/mm^2 on the top surface and below 300 thousand cluster/mm^2 on the bottom surface.



Quality doesn't otherwise seem to be impacted.

As of a month or so ago, we thought this was an issue with our cBot. But as of last week, tech support appears to be attributing all cases of this issue to reagents. Illumina is sending us extra reagents to compensate for the loss of reads. That is, we are running two flowcells to collect one flowcell's-worth of data.

Since both of our last 2 runs show this phenomenon, others must be seeing it as well. Anyone figured out a work-around that allows the normal cluster density to be restored?

--
Phillip