Uneven SYBR green staining/quantitative DNA gels - (Dec/05/2008 )

I'm trying to do some semi-quantitative PCR. I'm doing the PCR reactions, running them on a 1.2% agarose gel, staining with SYBR green, and quantifying the band intensities using ImageQuant. In initial runs to make sure that I'm in the linear range with my PCR's I noticed that bands near the edges of the gel were brighter than I would have expected. So I did some test runs to see if I could get an even signal. I ran identical 10 ul samples at various positions on the gel and then post-stained with SYBR green and did the quantification. Bands that should be identical are varying by more than 2 fold, with bands in middle lanes being dimmer than those in outer lanes. I'm subtracting background in the quantification so I don't think that uneven background is the source of variability in the signals. And I've ruled out the Typhoon imager I'm using as the source of the problem because these differences are apparent to the naked eye when I put the gel on a UV box. So those bands that should be identical either have varying amounts of DNA or varying amounts of stain present. I don't know why or how either of those things would happen.

I've tried the following, to no avail: increasing the volume of staining buffer so the gel was submerged deeper, increasing the time of soaking in the stain, adding SYBR green directly to the DNA samples before running the gel, and pouring the gel thicker. My next move will be to try a different gel box.

Any ideas/explanations?

Try adding the SYBR to the gel right after you make it, and cooled a bit (before pouring into the mould) Also, SYBR is light sensitive, one thing is after you make the gel try to cover it with something light-proof or keep it in a dark room. And when running the gel, try in the dark also. Only expose to light when you are loading and capturing the gel pix.

semi-quantitative is just that: semi. So I'd expect things like that, even with ethidium bromide. just improve the best you can and get on with it. if you really want accuracy, maybe you can try real time. which at times can be inaccurate also. haha