pooling genes in a solexa lane

[bioinfosm]

Hi,

This should be a common thing to do, pooling a couple of genes in the same lane of solexa flowcell, as it still gives appropriate coverage.

The assumption being, that 30bp reads would map uniquely. How much of an issue is it? Say there are a few 40bp regions shared by the 2 genes. Is it fine to pool them in a lane (no tagging or indexing) or better use separate lanes for each gene?

[apfejes]

We do whole transcriptome shotgun sequencing, where all of the expressed genes are pooled in a lane (or multiple lanes), and seem to get decent answers, so I'd imagine pooling two genes is pretty trivial. The answer comes down to a few simple things, related to what you want out of your experiment.

Are the genes so similar that you can't distinguish them, and do you care?

Are the genes from different patients/samples/libraries, and do you care that you won't be able to tell which each read came from?

If your aligner throws away multimatches (eg. Eland), do you care if there are holes in the genes (where there are regions of similarity)?

If your aligner can do multimatches, you won't be able to tell which one comes from which, so those regions will show higher coverage than the surrounding areas (if reads are mapped back to these areas from multiple locations), but do you care?

If you have snps in the regions of similarity, you might be able to look at the reads and determine the spanning sequences.

Mostly, this all comes down to what you're designing your experiment to uncover/prove. Without more information on what you hope to accomplish, there is no right answer to your question.

Cheers

[bioinfosm]

Thanks for the response
Yes we went through these questions, and I did some analysis on checking where those similarity regions fall.... turns out the genes are not similar in the 'interesting' regions, and mostly introns have the 200bp or so blocks of similarity.

just curious .. what kind of coverage do you shoot for .. using the whole transcriptome?
n i will be coming to your chip_SEQ tool soon ... so stay put

[apfejes]

That's cool - sounds like pooling is the answer, then.

For whole transcriptome, it doesn't make sense to as about "coverage". With a non-normalized library, coverage will range from extermely low for weakly transcribed transcripts (0-.5x coverage), to housekeeping genes and highly expressed genes (100+ times coverage).

I've only got a couple flow cells of transcriptome data at the moment, so I don't know the ideal coverage. That's still a work in progress.