Chip-seq sample prep protocols for 454?

[What_Da_Seq]

Hi everybody - first post - frequent visitor though,

I am doing some pilot work and I would appreciate some pointers towards sample prep protocols/papers for ChIP-seq intended for multiplex (barcoded) 454 sequencing in anticipation of the Titanium upgrade. BTW we will be outsourcing the sequencing so no inhouse expert to ask.

Any help is greatly appreciated.

Thanks

Frank

PS. Extremely informative site

[Chipper]

Why would you use the FLX for ChIP-seq? Even Titanium will only give 1 million reds per run. But you could look for the ChIP-PET papers in case you have to use pyro for this.

[What_Da_Seq]

If that is indeed the case - I was hearing something like 3Gbases but perhaps that quote was not per run. Sux to be on the bottom of the information flow.
And as I understand it ChIP-PET only assays multiple fragments per read. Since we are interested in genomic DNA and not cDNA that does not help.

Gotta check on that 3GB figure though. Word of mouth and a faulty central processor (aka my brain) is not a good combination.

[Chipper]

Ok, the last figure I heard was 0.5 Gb. But the thing with FLX is that you get longer reads than the other methods so even 3 GB is not much compared to SOLiD or Illumina in terms of tag numbers, which is what counts for ChIP seq. I am not saying it will not work, it is just going to be more expensive. If you intend to barcode it will not leave many tags per sample.

The idea with the chip-pet protocol was to generate more uniqe fragment ends by concatenating 18-mers (?) and was used with 454 in a few publications (cMYC and I think also p53) before the Solexa was released.

[What_Da_Seq]

Just stick with me

3Gb genome. 95% eliminated in ChIP leaves 150Mbases. 0.5Gbases/ run gives you already 3x coverage without multiplexing though. The length of the seq run is exactly what is desired - no averaging out of multiple potentially independent events in 200bp stretch. Appreciate the input nevertheless.

Frank

[Chipper]

Ok, if you have a chip protocol that does not give you any genomic background I guess it would work... I agree it would be nice to be able to read the whole fragments to get better mappings and to get both end positions, but it will not be much different from doing paired ends with shorter fragments. Also, to get a resonable resolution with so few reads you will have to sequence shorter fragments so how are you going to use the full capacity (0.5 G bases is for 4-500 bp reads)?

What do you mean by averaging out independent events?