Size Bias in Minipreps - Which method is the best? - (May/23/2012 )

I am doing high-throughput sequencing of a library of barcoded plasmids in E. coli. The experimental plan is like this:

-grow two populations of cells, one control and one experimental (say, adding a certain metabolite)
-isolate plasmid DNA from each population
-mix and multiplex sequence the barcodes using Illumina HT-seq

The basic idea is that the prevalence of certain barcodes indicates that cells containing those plasmids are more prevalent in the population, and therefore have a faster growth rate.

It's therefore important that the population of the cells be accurately represented in downstream purification steps. For this reason, at each step the DNA should accurately reflect what the population of cells was like. This also means that the protocols used to isolate, purify, and amplify DNA not have any size or sequence bias.

My library currently consists of members anywhere from 4kb to 11kb, but mainly between 4kb and 7kb.

What methods are best for minimizing size and sequence bias? In our lab we use Qiaprep columns from Qiagen for routine work - does this have a size bias in the 4kb to 7kb range?

Does alkaline cell lysis followed by ethanol precipitation have a size bias?

Are there any other methods that would work for this application?

Thank you!

Rather than relying on not having size bias, why not make the plasmid sizes identical except for a small bar code difference? Then you would not need to worry.

Thanks for the thought - unfortunately unlike a phage display library the things I'm screening have lengths ranging from 800bp to 8kb. Any thoughts?