massive multiplex PCR-library for 454-sequencing

[CG&R]

I could really use some feedback on the following:

My situation:
- I have 400 genomic DNA samples.
- They contain 60 highly-similar regions of interest (300 bp).
- I can amplify them all with just 3 PCR primer pairs: giving me a mix of
20 products (300bp) in each PCR reaction.
- I use different barcodes for all the PCR reactions and have extended my
oligo's with KEY and Primer A (forward) and B (reverse) sequences of 454.
- I do every PCR reaction twice, with different barcodes.

This gives me 2400 uniquely barcoded PCR-products of the same length (420bp). I would like to use one half/full titanium run on a 454.

I know that equimolar pooling is going to be tricky.

Now my question:
- Should I just mix all PCR products together, take a sample, do clean-up, sequence it and hope for the best?
- All PCRs that I have done will be quantitative, so I have end-point fluorescence for every sample: should I try and do equimolar pooling based on this value and then clean-up.
- or should I clean-up every sample separately, then use picogreen to measure concentration, then do equimolar pooling, and sequence?

Anyone have experience with >96 barcoded samples and how to do equimolar pooling, did you get extreme bias or was it doable?

Any feedback/suggestions and ideas are much appreciated!

[Corey]

Thoughts about normalization

The Invitrogen SequalPrep has worked well for us for 96 samples. Not cheap, perhaps you could do a rough normalization then pool 5-10 samples/sequalprep well?

Corey

[watomar]

Hello CG&R,

In my experience, QPCR is the most accurate way to quantify fragments prior to pooling. It is definitely going to be more accurate than pooling from pico results but if running QPCR on that many samples isn't an option for you than you would benefit most from removing liquid handling variations in your assays.
For pools of 96, CVs of 3-6% are pretty good. This corresponds to a ratio of about 1.4 of your most represented sample to least represented sample in the pool.

[CG&R]

Thanks for the input!

Currently we are setting up the following protocol:

4x96 samples x 3 exons x 2 separate PCRs = 2304 PCR products

Since we perform this as a q-PCR (EVAgreen) we should have some indication what the relative amount of product is in every reaction (ruling out primer-dimers atm). For every exon we pool 8 samples with relative equal amounts of DNA and use 25µl for a sequalprep plate.

In the end we will have 3 x 96 sequalprep plates that should give 20µl per well so about 6 ml of PCR product mix, of which we will use only a small amount for 454-emulsion PCR

Will let you know what our results will be.

Cheers!