[dvh]

first experience with nimblegen seq cap on GAII

I wish to withdraw the post (#47) above.

There was a sample mix up on the flowcell. Lane 1 was actually lane 8 etc. Found this out from the PhiX control, the fact we had mixed sample types, and careful inspection of the data. Worth pointing out to the list that it is VERY easy to flip the 8-well tube of samples (or perhaps flowcell) in the cluster station.

To apologise for the earlier post, here are some results from the first lane we've looked at. Single end solexa reads. We have only very crudely dealt with the (unwanted) adapter sequence used by nimblegen for PCR amplification at present, much more optimisation is possible which will improve both number of reads and quality of reads aligning. Using maq map -n 3 to the whole genome, 1.3m reads map with MQ 70-99 i.e. uniquely with high confidence. Of these 85% map to the regions we selected for the seq cap chip. So it works. Somewhat variable coverage, I'll try and post some stats once we've done some more analysis.



david

[dvh]

first experience with nimblegen seq cap on GAII - ?problem


Edited 10 SEPT 2008: I wish to withdraw this post - see post 48 below:

Hi,
We've just run our first nimblegen seq cap service (i.e. they did the probe design/hyb/elute/pcr amp steps) sample on a solexa flowcell. We confirmed the returned sample looked OK on bionanalyser.
No problems with solexa library prep.
Flowcell worked well (GAII) for other libraries.
After removing the 454 adapters, almost no reads map to the regions we asked for on the chip (custom design 1.5Mb). However the reads map very well to the human genome (maq alignment to whole genome, >90% map, error rate 1.0%). But to a few distinct regions which look like they might be repeat regions.

We have some other samples to run, and will repeat the library prep, but am a bit concerned by this. Wondered if anyone else has had any difficulties?

david

[sigusn]

Some references>
Okou et al. Nat Methods. 2007 Nov;4(11):907-9
Excelent method description in the supplement

Hodges et al. Nat Genet. 2007 Dec;39(12):1522-7.
Describe the sequence capture followed by Illumina Genome analyzer

The custom made sequence capture arrays from Nimblegen now come with full protocol for the sample prep, should work for Illumina if you just use the right oligos.

[rosatoc]

My understanding is that a custom seq cap array from NimbleGen runs around $1K. As we prepare to offer a seq cap in this core facility I'm glad to know there is a thread I can come to for tips on technical points. Is there a reference for the PCR amplification post elution (Okou et al?)? Since the initial expectation for sequencing seq cap material was for the 454, I'm pleased to know that seq cap is working with the Illumina (the one platform we have). Thanks!

[sigusn]

pooled DNA

Yes, I have tried pooling 4 samples and use the Nimblegen Seqcap arrays. With average sequence coverage of 40-50X I think I can say that works OK.

[fgibson]

DNA pooling with nimblegen sequence capture arrays

has anyone tried using pools of genomic DNA on the nimblegen sequence capture arrays?

[sigusn]

Hi
I hybridized for 65h. After elution I used speedvac to dry down the sample, then dissolved in water. I did not get as much as Okou et al but enough to perform the PCR step and then the PCR product was seen on gel and on bioanalyzer. This was true for 4 out of 6 samples i worked with, I have not found out what happened to the two that gave no PCR product. Probably low quality of the input DNA.
I would guess that the elution product is single stranded DNA (the Nimblegen arrays are only designed on the forward strand) and single stranded DNA is not seen on agarose gel with EtBr staining.

[lambroso]

Hi Sigusn,
I did the elution similiar to your method. I placed the bottom piece of a Corning hybridization chamber on a piece of tin foil in a 100 degree heating block, with out the blocks in it. Then put my slide on it and added 2mls of heated water. I know 2mls sounds like alot, but I found that it kept the slide covered for the total elution time and could be gently mixed without spilling off. I used the hyb chamber in case anything spilled off the slide. Lucky for us we have a good speed vac. But, the bottom line is I still haven't gotten this to work yet as verified by running the speed vaced elution on a gel. I started out with 25ug and if I had similiar results as Okou, who recovered .7-1.2ug, I should have been able to see it on a gel. How long are you eluting for? I'm not sure what to try next. Are you hybridizing for 60 hours? I was just doing it over night but may need to increase it. Also curious about what Nimblegen's special unit for elution is like.
Thanks again for you thought on this.

[sigusn]

Elution step

Hi Lambroso
The elution step is well described in the Okou et al. paper. (Nature Methods 2007). However i did not have any equipment from Nimblegen for this. So I used home made system, cutout aluminum block, for holding microarray slides and silicon rubber (Elastosil RT 601 A/B, Wacker-Chemie GmbH, Munich, Germany) that I cut out to fit the slide to minimize the area that I pipetted water on. Then placed this on a heat block and hoped it would stay at 95°C wile I pipetted 400 ul H20 on the slides three times. Of course 1/4 of the water evaporated before I could remove it from the slide, but that should not be a problem. I just received my next batch of slides from Nimblegen, and now they come with a special unit for the elution step. How do you plan to perform the elution?

[lambroso]

Hi Sci Guy and Sigusn,
I am also trying to do a seq. capture/GAII approach similiar to the Nimblegen arrays using a cdna array. Sounds like I'm doing something similiar to Sigusn, as far as using 100-300bp sonicated dna and following Illumina's prep for end repair/adaptor ligation. Next step is to hybridize to a cdna array, elute and then continue with Illumina's protocol. I wondering how you did the elution step. Thanks for your help.

[sci_guy]

We are bound by their standard protocol I think. The service specifies delivery of 21 ug of DNA and they do the library prep and QC. The concatamerisation/shear/ligate Illumina adaptor method post-capture seems very messy from an in silico point-of-view if you consider that a high proportion of the reads will contain adaptor sequence starting at an essentially random location within a read; particularly so for mutation discovery.

[sigusn]

Yes Im capturing 6 separate contiguous regions total 1.25 Mb. If you are doing the seq-cap as a service at Nimblegen, will you have the option of using shorter fragments with Illumina adapters, or do you have to go with their standard protocol of 500 bp with 454 adapters?

[sci_guy]

Excellent! Thank you Siqusn. Those numbers are informative for us. One last question, are you capturing a contiguous region (less Windowmasker repeatmasked regions)?
We are capturing a contiguous 5 Mb region. Nimblegen suggest they have about 70 - 80% specificity with standard ~500-mer library preparation and tiling over a large area.

[sigusn]

Hi
Hodges et al write “For optimal sequence generation on the Illumina platform, fragments of less than 250 bp in length are desirable. This was below the size range initially optimized for capture. ” But hen they could optimize the size range, they don’t mention that?
Im not sure how Hodges et al define “specificity of the capture” they write. “…found a reproducible threefold decrease in the specificity of the capture with shorter fragments (average size of 100–200 bp )(Table 2) However, this was more than compensated for by the increased sequencing efficiency and the broader distribution of fragment ends ”
What I can read from the tables is that in Table 1 they get from 100 K to 2 million reads per array and about 30-55% of the reads map to the selected exons. In table 2 they get 1,85-2,3 million reads and around 30% of the reads in selected regions.
In my data, I get 4-5 million reads per sample (lane) and the fraction of sequence reads that align to the captured region varies between 32-56% (same array different DNA samples).

[sci_guy]

Hodges et al. reported that following something approximating your method they could get good Illumina GA-1 reads but suffered from poor sequence capture, whereas 500-mers could be captured OK but suffered from poor numbers of reads. Do you note sequence capture was less than optimal with your 100 - 300mer library?