[Methods-and-reagents] Re: sizing DNA fragments Macintosh

Austin P. So (Hae Jin) nobody at nowhere.com
Tue Jun 28 23:51:36 EST 2005


Thomas Isenbarger wrote:
> In article <d9suej$fa3d$1 at netnews.upenn.edu>,
>  kjude at mail2.sas.upenn.edu (Kevin M Jude) wrote:

>>It can be done with any plotting software; prepare a standard curve and 
>>see where your bands fall.  Migration rate increases logarithmically with 
>>decreasing size.

> but, i have tried this, and i am looking for a quicker solution.

I don't think there is a "quick" solution...unless you are absolutely 
sure your buffer composition, your gel percentage, your current, and 
your temperature of running conditions...DNA will even "smile" across a 
gel, so that migration distances are not uniform...

> i plotted the data in igor, then fit 1/distance versus size with a 
> second order polynomial.  then i used the fit equation and the distances 
> of the unknowns to solve the quadratic equation for the sizes of the 
> unknowns. . .and my control plasmid band was 0.5 Kbp off.

Your mistake was plotting 1/distance. You are introducing a big error in 
your fit.

A reciprocal plot is nowhere near the same as a logarithm fit BTW...why 
on earth did you do this?

*Always* fit to the raw data. And make sure your standards fall within 
the resolution capabilities of the gel and that your unknowns fall 
within the range of your standard curve.

Austin

> so, i am looking for a quicker method with better fitting to do this.  
> there are published fitting algorithms that are better than a simple log 
> fit, so i am hoping to find software that has these implemented.
> 
> when you say "see where your bands fall", are you saying just eyeball 
> the place on the curve where my unknowns fall?
> 
> this is a bit too rough for me.
> 
> any other help?


More information about the Methods mailing list