Bridge amplification and fragment size

[BTS]

Hi All,
I'm new to this whole NGS business. I will be using an Illumina GAII. In a recent article from Bio Techniques, I read that longer (about 600bp) fragments can increase the uniformity of coverage across a targeted region. In the paper, amplicons were fragmented to 200bp prior to adaptor ligation and subsequence sequencing. Does anyone know why this would increase uniformity? Also, does anyone know if it is possible to successfully generate clusters during bridge amplification with fragments larger than 200-300 bp? And if so, how large?

[ScottC]

Illumina tends to recommend up to 600b or thereabouts.

[aperera]

At a recent Illumina user meeting, we were told the ideal insert size is 150-200bp.

[ScottC]

That's possibly the ideal size, but it works with larger fragments than that.

[huguesparri]

I used PE fragment as high as 800 pb (including adaptors) and it did work fine.

[aperera]

That's right, it is the ideal. We have used larger sizes as well.

[dmborek]

Are there any negatives of using longer inserts e.g. reduced number of clusters due to their larger size?
Anyone used even longer than 800 bp for PE (not mate pairs)?

[konrad98]

Beyond 800bp you start getting cluster sizes which are almost guaranteed to merge with each other during the course of a run. You can can get around this by lowering the cluster density to around 100-200k/mm^2

[kmcarr]

Besides cluster size as konrad noted there is also an issue of cluster shape. Due to the randomness of the direction which the "bridge" may form relative to the anchor point of the strand, clusters may start out being somewhat misshapen, e.g. oblong, jagged, etc. It takes a certain number of amplification cycles until the shape of the cluster becomes round. This shape effect is exaggerated with longer inserts, and for very long (800bp) inserts you may need to increase the number of amplification cycles. Of course additional cycles will make the clusters even larger.

The shape of the cluster matters because the image analysis algorithms look for round spots only. Odd, misshapen spots are rejected. The size of the cluster may also have an effect on cluster identification. Spots which are too large may be rejected by the software as potential artifacts. You really should discuss this with your Illumina FAS.

[dorix]

Hi,
I have libraries of insert size of about 600bp. I'm quantifing them by qPCR. Those are amplification free libraries, so I gess I may have same fragments wihout adapters and so on. Unfortunatelly comparing molarity from bioanalyzer and from qPCR are incosistent: qPCR shows two times lower amount. I'm wondering which is more likely: I have samples with not fully constructed fragments or amplification efficiency is that much lower with longer fragments. I'm affraid of overclustering, using results from qPCR, however I do not want to loose money by less density.

Does anyone has similar expirience?

[nahmed]

Time required to sequence one base

Hello,
I am Nauman Ahmed. I am a new new member of SEQanswers.com. I am a PhD student in TU Delft.
My question is general about the time to sequence one base in Illuminia SBS technology, but let me give an example: In the specification sheet of Illuminia HiSeq 4000 , the run time is 3.5 days at maximum, generating 2 X 150bp reads. Does it mean that to sequence one base (one cycle), it takes about 17 minutes?

[misterc]

Hi Nauman,

On HiSeq 4000, its about 16 minutes per cycle but this is limited by the imaging time to excite the incorporated fluorophores and image them simultaneously. The chemistry time is about 4 minutes but since it runs 2 flow cells simultaneously, the cycle time is about 8 minutes per side. If you ran 1 flow cell by itself, then the entire run time would be reduced for 2x150 cycles as you'd then have a 12 minute cycle time instead of 16. Hope this helps.

Best,
Charles

[Jessica_L]

Hi,
I have libraries of insert size of about 600bp. I'm quantifing them by qPCR. Those are amplification free libraries, so I gess I may have same fragments wihout adapters and so on. Unfortunatelly comparing molarity from bioanalyzer and from qPCR are incosistent: qPCR shows two times lower amount. I'm wondering which is more likely: I have samples with not fully constructed fragments or amplification efficiency is that much lower with longer fragments. I'm affraid of overclustering, using results from qPCR, however I do not want to loose money by less density.

Does anyone has similar expirience?

Hi dorix,

The issue you're having is, I think, rather common. It helps to look at the overall shape of the electropherogram in the bioanalyzer data and make sure that it looks like appropriate for your library type. Illumina manuals usually have examples of what good/passing bioanalyzer data looks like. See page 25 of the PCR-free manual for an example of what I mean.

The other thing to consider is the quality of the qPCR data. If the standard curve looks good and passes all the necessary quality criteria (i.e. difference in Ct values between data points on the standard curve) and the efficiency and R^2 values are within the acceptance criteria, then I'd probably lean toward using the qPCR data over the bioanalyzer data. As you've mentioned already, there's a good chance that some of the material in the bioanalyzer traces will have adapters in an orientation that will not lead to successful sequencing, so the qPCR data gives you a better idea of how much material will bridge amplify and produce clusters.

I hope that helps.

[nahmed]

Hi Nauman,

On HiSeq 4000, its about 16 minutes per cycle but this is limited by the imaging time to excite the incorporated fluorophores and image them simultaneously. The chemistry time is about 4 minutes but since it runs 2 flow cells simultaneously, the cycle time is about 8 minutes per side. If you ran 1 flow cell by itself, then the entire run time would be reduced for 2x150 cycles as you'd then have a 12 minute cycle time instead of 16. Hope this helps.

Best,
Charles

Thank you for the reply . If I look at the specifications of HiSeq 2500, the sequence runtime in upto 6 days (in high output mode) which means that it has much longer cycle time as compared to HiSeq 4000. It generates smaller reads i.e. 2x125. Is it due to older technology or something else?
http://www.illumina.com/content/dam/...-portfolio.pdf

[dorix]

Hi dorix,

The issue you're having is, I think, rather common. It helps to look at the overall shape of the electropherogram in the bioanalyzer data and make sure that it looks like appropriate for your library type. Illumina manuals usually have examples of what good/passing bioanalyzer data looks like. See page 25 of the PCR-free manual for an example of what I mean.

The other thing to consider is the quality of the qPCR data. If the standard curve looks good and passes all the necessary quality criteria (i.e. difference in Ct values between data points on the standard curve) and the efficiency and R^2 values are within the acceptance criteria, then I'd probably lean toward using the qPCR data over the bioanalyzer data. As you've mentioned already, there's a good chance that some of the material in the bioanalyzer traces will have adapters in an orientation that will not lead to successful sequencing, so the qPCR data gives you a better idea of how much material will bridge amplify and produce clusters.

I hope that helps.

Thank you for your response, Jessica. The shape seems fine to me, Gaussian type shifted towards higher lengths. I have also seen amplicon libraries without amplification, which has given me a sense of libraries construction yield (that it is not very efficient). Nevertheless, so happend I've tested other qPCR kit with shorter amplification steps and it has given me even lower results, with comparable for short libraries. That warned me even harder :/

I wander if anybody have an expieriece of overclustering with that kind of observation?