HiSeq questions

[jo_mason]

Just a few quick question...
How many of you have a HiSeq2000?
How many of you actually have a working HiSeq2000?
If you have a working HiSeq2000, how long did it take Illumina to install and validate the machine?

We have a very long story that I won't bore you with...but it took 6 months from ordering to shipping, it arrived and was unpacked on 4th Jan, and we still haven't had a single successful run out of it! It officially scrapped through validation last week and promptly broke down again. I'm trying to work out if they dropped it on the way, or if this is normal?

[NextGenSeq]

Ours worked first time and is still working great.

[cascoamarillo]

We have the HiSeq 1000 installed last week. This week comes the validation run; if everything works fine, next week the training run.

[westerman]

Our HiScanSQ literally went up in smoke during the first validation run. The unit was replaced and syringe broke during the second validation run. The third run is almost complete and I expect to be working on the analysis tomorrow. All complex low-unit machines can have problems. The major issue is how fast and well does the manufacturer fix the problems. So far we are satisfied.

[csc]

Installation took about 3 days. Validation took about a week. We have had some problems with it so far but they have been mostly fluidic - I think all were clogged or broken solenoids. I don't think both flow cell positions were ever down at the same time so we've been able to keep running.

[rabbitstew]

We have many HiSeqs and they've been hit or miss. The first batch that we got a year ago were smoothly installed and run with few problems. The later shipments have had varied issues, with one machine that can not validate and has been through multiple FSEs (as in multiple people coming to try to figure out the problem, not just multiple checks). We had some really basic problems (syringes not tightened, platform not level, etc.) and some trickier ones (complete reverse flow, major clogs on a brand new machine), and some have worked just fine from the start.

Perhaps these quality discrepancies are a product of high demand and trying to get them out asap?

[NextGenSeq]

Illumina did have a bad batch of flowcells where the edge of the lanes didn't have oligos deposited correctly. This showed up as a black stripe with no clusters at the edge. Supposedly this has been fixed.

[jo_mason]

We have many HiSeqs and they've been hit or miss. The first batch that we got a year ago were smoothly installed and run with few problems. The later shipments have had varied issues, with one machine that can not validate and has been through multiple FSEs (as in multiple people coming to try to figure out the problem, not just multiple checks). We had some really basic problems (syringes not tightened, platform not level, etc.) and some trickier ones (complete reverse flow, major clogs on a brand new machine), and some have worked just fine from the start.

Perhaps these quality discrepancies are a product of high demand and trying to get them out asap?

That's a really good point and niot something we had considered.....we were thinking perhaps it was geographical. We now have access to 3 HiSeq's, all recent acquistions, and not one of them is working. Two have been with us for 4 months, 1 only for 1 month. The problems have been one thing after another, and all simple things like you said. We have also not been helped by the awful communication that Illumina and their distributor in Malaysia seem to suffer with.

[shruti]

Hi, was wondering if anyone can tell me what is the expected throughput in Gb, if I run 2x36bp , 2 lanes for human whole genome in hiseq2000.

[greigite]

We had a HiSeq 1000 installed recently. Validation run went fine. Midway through read 3 on the training run (read 2 = indexing read) the hard drives died completely and the whole system stopped. Before that there seemed to be a fluidics problem causing one lane to disappear entirely on read 3, after showing good cluster density on read 1. These two problems seem totally unconnected.

[jo_mason]

Hi, was wondering if anyone can tell me what is the expected throughput in Gb, if I run 2x36bp , 2 lanes for human whole genome in hiseq2000.

6.5-8.75 Gb from 2 lanes of a HiSeq flow cell running 2x36bp.

[monad]

Another way to look at the HiSeq output is 80-100M passing reads per lane. You can multiply with your sequencing reads to get your Gb. For Example, with 2x100 you can expect about >16Gb output if you hit the optimal clusters.

[NextGenSeq]

No for two lanes you should get closer to 15GB.

Good libraries should give 100 million clusters per lane. This is supposed to double this summer so soon you can get 30GB.

[monad]

Today's $5k per genome is based on improved chemistry for 600Gb. This output seems based on 16 channels (two entire flowcell) at 2x100 at ~190M cluster density. My realistic expectation based on differentials between promising and deliverable, I expect about 160-180M, or 32-36Gb at 2x100 per lane.

This output is projected to be up again later this year about 60% higher as they projected early this year at AGBT.

[HMorrison]

We had a HiSeq 1000 installed recently. Validation run went fine. Midway through read 3 on the training run (read 2 = indexing read) the hard drives died completely and the whole system stopped. Before that there seemed to be a fluidics problem causing one lane to disappear entirely on read 3, after showing good cluster density on read 1. These two problems seem totally unconnected.

They may be totally unconnected, but we had the identical two problems on our training run (HiSeq1000 here too)--lost half of read 3 when hard drive failed, but before that we lost lane 1 due to solenoid/fluidics issue. Still waiting for data and struggling with OLB (will start new thread for that).