Illumina Human Exome vs Agilent Human Exome

[GW_OK]

I know it hasn't been out for very long but perhaps we have a few early adopters in the audience who might care to weigh in on these two whole exome capture systems? I particularly like the Illumina internal control idea, plus the fact that it supports the new, longer read lengths. And given the price afforded by pooling it seems to be the new way to go. Agilent, though, says their baits are better, longer, and more numerous for increased on-target coverage.

What say you, SEQanswers?

[upenn_ngs]

Although the pooled enrichment is less expensive per sample, the increased targeted region (62Mb illumina vs. 38Mb agilent) means more money will be spent on sequencing per sample. For example, the illumina product guide shows ~80% at 20x coverage for 6 pooled samples; however, this data was collected from 5 lanes on HiSeq.

[ntremblay]

The SureSelect kit is by far the reference right now, most of the resequencing project have adopted it. With the new 50 Mb version, this kit has increased it's coverage of all exons annotated in the consensus CDS (CCDS – March 2009) database as well as 10 base pairs of flanking sequence for each targeted region. In addition, the content contains small non-coding RNAs from miRBase (v.13) and Rfam. Another great feature of this kit is that it comes with validated protocols for Solid and Illumina technologies.

we've been using this kit so far and we're more then happy whit it. Also, there's no limitation in terms of sample multiplexing... so on the Solid 4 we've been pooling 8 sample with a average coverage of 30X witch is more then enough for snp discovery.

As for the TruSeq kit from Illumina, it has an almost identical coverage plus some other feature included in the capture ... that makes the 62Mb of capture. The advantage of this kit is the low price per sample to do the capture (about 300 vs 750) but you'll end up sequencing more on a GaIIx to get the same amount of data as on a Solid with SureSelect...

bottom point, I'm sure both kit work well and will provide you with a excellent capture tool for exome resequencing,

cheers,

[frozenlyse]

I'm not sure why capturing more of the genome would be a disadvantage, unless the extra regions captured in the Illumina version are junk? Sure it's going to take more sequencing to get the equivalent times coverage, but won't you find more mutations?

[ymc]

I am thinking about trying exome sequencing on Illumina's MiSeq.

MiSeq's max throughput is 1Gb in 27 hours with 75% bases >= Q30.

If I use Agilent SureSelect, the coverage will be about 26.3x. Is this good enough to have an error rate of at 10^-6 per base???

What is the state-of-art coverage calculation? I find Churchill & Waterman's old paper:

http://www.cmb.usc.edu/papers/msw_papers/msw-107.pdf

If I assume the average error rate to be 1 in 100 bases (I suppose that's not far fetched given 75% bases >= Q30), the Churchhill & Waterman model gives me an error rate of 1.36x10^-10 for 26.3x.

Alternatively, if I use Illumina's kit, then the coverage becomes 16.1x and the error rate 5.37x10^-7.

Both give me acceptable error rate. But I suppose there is a better way of calculating this now???

[dottomarco]

the new SureSelect targets 50Mb, why would you use the old version (38Mb)?

Sequence enrichment usually yield 60/70% of reads on target... and just 75% of these will be HQ... so you end up with 10X coverage (with HQ reads). That's quite low

I thing the MiSeq is more likely to be used for smaller target resequencing or amplicons (it was in fact designed for diagnostics). To be able to get a HQ exome you would need at least a 3Gb throughput, in order to get a usable 30X.

[upenn_ngs]

from our experience, a minimum 4-4.5 Gb needs to be collected for the Agilent exome. it does not appear that MiSeq was designed with exome resequencing in mind; but certainly a smaller targeted capture would be appropriate.

[NextGenSeq]

We are runnning several Illumina whole exome lanes now on the HiSeq. Once I analyze the data I'll post the coverage here.

One thing I don't like is that the protocol is very long and tedious. You do two rounds of amplification. Also, you first use the TruSeq DNA kit to make the initial library before the whole exome capture. In our hands about half the libraries didn't yield enough to continue on with the whole exome capture (you need a minimum of 500 ng).

[DMO]

We have tried it out as well. I agree with NextGen about the prep, it is rather long and tedious. We are still sequencing our first batch, so not sure how the data looks yet.

[ymc]

from our experience, a minimum 4-4.5 Gb needs to be collected for the Agilent exome. it does not appear that MiSeq was designed with exome resequencing in mind; but certainly a smaller targeted capture would be appropriate.

Thanks for your reply. So what is the error rate per base you can achieve with such coverage???

[NextGenSeq]

I was told that they are targeting the clinical sequencing market. The MiSeq automates the library and cluster generation using the EpiCentre Nextera library reagents. Apparently thats why Illumina bought Epicentre.

[ymc]

I was told that they are targeting the clinical sequencing market. The MiSeq automates the library and cluster generation using the EpiCentre Nextera library reagents. Apparently thats why Illumina bought Epicentre.

EpiCentre says they are working on making Nextera works with Agilent SureSelect system. If they are successful in making it work, will that mean much lower DNA input, shorter prep time and no Covaris needed??

[DMO]

Yes. The insert size for the HMW buffer with Nextera is much larger than the target insert of 175bp for the Agilent SureSelect protocol. This is going to affect the efficiency of the hyb and the resulting pull down. If they get it right, nextera+ Agilent targeted capture+MiSeq will be the go-to combo for Clinical Sequencing.

[krobison]

The 454 protocol for SureSelect shows a fragment size peak close to 700 bp; anyone know if specificity is lower with the 454 kit?

Of course, 454 has long reads -- with paired ends, you do run a bigger risk of your bait correctly capturing an exon in the middle of the fragment but your reads mostly covering neighboring DNA.

[starlight312]

Yes. The insert size for the HMW buffer with Nextera is much larger than the target insert of 175bp for the Agilent SureSelect protocol. This is going to affect the efficiency of the hyb and the resulting pull down. If they get it right, nextera+ Agilent targeted capture+MiSeq will be the go-to combo for Clinical Sequencing.

I'm might have this wrong and I'm not too knowledgeable when it comes to the numbers but I'd like to ask, what is the insert size range optimal for the HMW buffer for Nextera? I ask this because the current optimal insert length for the HiSeq is from 200-500bp. Assuming that the MiSeq is using identical chemistry w/ the HiSeq, then either way, we can't have the insert sizes too large (800bp+) because that's when it gets really messy.

Thanks =D