[Lee Sam]

200 million reads per lane? Is that the standard or are you pushing the limit? What is the standard # of reads/lane for a HiSeq?
Is the RTA of the HiSeq the same as GAII? In a GAII run the low-quality ends of reads are flagged with a quality score of 2 (ASCII B). Does the HiSeq RTA also output reads in this manner? What % of reads in the first and second run of a pair are low quality (QV=2)?
Is it comparable to GAII (<15% in first run, up to 40% in second run).

Sorry, I should have been more clear. 100 million paired reads (or 200 million ends).

[NextGenSeq]

We've been waiting for six months and it's still not here (a month after it's scheduled delivery).

[jo_mason]

Can anyone tell me how long it took for Illumina to commision your HiSeq after delivery. I've just had a timetable sent to me which takes a little over two months and that doesn't include the training!

[huguesparri]

We "routinely" generate 70.000.000 passed filter sequences per lane on our HiSeq2000.

[juan]

200 million reads per lane? Is that the standard or are you pushing the limit? What is the standard # of reads/lane for a HiSeq?
Is the RTA of the HiSeq the same as GAII? In a GAII run the low-quality ends of reads are flagged with a quality score of 2 (ASCII B). Does the HiSeq RTA also output reads in this manner? What % of reads in the first and second run of a pair are low quality (QV=2)?
Is it comparable to GAII (<15% in first run, up to 40% in second run).

[Lee Sam]

Hi mgogol

As far as the propriety pipeline of Illumina are concerned they remain to be same but in our experience,we have seen significant increase in the processing times as the raw data goes up 4-5x / lane on HiSeq. If you are thinking about buying IT infrastructure I would definitely consider having 1-2 machines with quad cores with > 64GB Ram if possible. Although 32 gb is feasible but as the cluster density on the HiSeq flowcell goes up chances are you will need more ram.

Hope this helps,
-Abhi

We've seen a similarly huge increase in the processing times. My personal experience: aligning ~180-200 million 2x100PE whole-genome low coverage data reads from single lanes with BWA from *_sequence.txt files gives align (bwa aln and sampe) times in the 16-20 hour range per lane on 8 processors, not including downstream analysis.

[mgogol]

Thanks, that's helpful.

[apratap]

Hi mgogol

As far as the propriety pipeline of Illumina are concerned they remain to be same but in our experience,we have seen significant increase in the processing times as the raw data goes up 4-5x / lane on HiSeq. If you are thinking about buying IT infrastructure I would definitely consider having 1-2 machines with quad cores with > 64GB Ram if possible. Although 32 gb is feasible but as the cluster density on the HiSeq flowcell goes up chances are you will need more ram.

Hope this helps,
-Abhi

[mgogol]

Can anybody with a HiSeq comment on their computational resources downstream?

I see the site prep guide, do you have systems like this?

http://seqanswers.com/forums/attachment.php?attachmentid=461&d=1280960975

[bioinfosm]

Thanks for your reply, but I am a bit confused- what do you mean by topped off in between runs? Also, do you mean that each individual flow cell on the HiSeq uses the same amount of reagents as a single flow cell on the GAIIx, or that both together use the same amount as a single GAIIx flow cell (so effectively half the amount of reagents per flow cell)?

Sorry I did not check back earlier.

James might have already answered you. I will be at a webinar on hiSEQ next week and will add any updates here

[james hadfield]

The GAIIx and HiSeq use essentially the same amount of reagent per flowcell. However the HiSeq can run two flowcells at once and each flowcell generates 3-5x as much data.

The word 'run' does not actually tell you very much about what someone is doing. It has always been a bit of an issue when comparing SOLiD to GA as one had two slide where the other used one flowcell.

I find it much easier to describe the run in terms of it being single or paried end and by stipulating how many bases are being sequenced per read; e.g. SE36, PE72, etc. This would always refer to a single slide/flowcell.

[snetmcom]

right now the GAII's are running more often than the HiSeq. $$$ and we are having what looks like a weird reagent issue. Not to say it is reagent based yet, but straaange.

[greigite]

I think it uses the same amount, just that one can load much more at the beginning, so it does not need to be topped off in between the run

Thanks for your reply, but I am a bit confused- what do you mean by topped off in between runs? Also, do you mean that each individual flow cell on the HiSeq uses the same amount of reagents as a single flow cell on the GAIIx, or that both together use the same amount as a single GAIIx flow cell (so effectively half the amount of reagents per flow cell)?

[bioinfosm]

I think it uses the same amount, just that one can load much more at the beginning, so it does not need to be topped off in between the run

[greigite]

We have our first HiSeq here at IGS but it will take 2-3 weeks before we will have production data. For now there will be some testing runs.

What I do know is, the final output is a BCL file from two flow cells, binary equivalent of qseq which can be converted back to qseq using a Illumina supplied converter. The downstream analysis from that point is same as of GA.

Hope this helps!
-Abhi

Abhi, Could you comment on whether the HiSeq uses the same amount of reagents per run as a GAIIx or 2x as much since it is running 2 flow cells in parallel?