mRNA level and protein level correlation?

Ian A. York iayork at
Sun Mar 21 22:57:37 EST 2004

In article <h0pj50hqbcau2r58fiks7c8u4ehdlas26o at>,
Peter Frank  <peter_frankde at> wrote:
>Michael L. Sullivan wrote:
>>Minimally, I think one needs to run a dilution of each sample, and 
>>probably run multiple blots to get believable numbers.
>I haven't done dilutions (yet) but I have done multiple blots and
>technical replicates on each blot. Then I evaluated the data
>statistically. I am still not very confident in the absolute value of
>the fold change but now I am at least confident enough about the
>direction and the significance of the change.

I agree with Michael about the problems with westerns, and that using 
internal dilutions are the only (non-Rube-Goldbergian) way of getting 
semi-quantitative Westerns.  Even doing it very carefully, I find it's 
very hard to be confident about two-fold differences in protein 
concentrations (my steps in western blot dilution series are usually 
three-fold for that reason).  

Another possible way to measure your protein is by radiolabelling and 
immunoprecipitation, if you have antibodies that will work in IPs (or if 
you can make them work in IPs [1]).  While immunoprecipitation measures 
synthesis rather than accumulation of protein, a long label can 
approximate accumulation, and IPs are much more quantitative than are 
westerns.  There are still technical challenges in a long label, of 

In general from your description I'd be inclined to agree with you, that 
the problem is more likely one of accurate measurement than of biology.  


[1] If an antibody only recognizes denatured antigen, as in westerns, you 
may still get it to work in an IP by denaturing the sample first, e.g. by 
heating or adding SDS.

    Ian York   (iayork at  <>
    "-but as he was a York, I am rather inclined to suppose him a
     very respectable Man." -Jane Austen, The History of England

More information about the Methods mailing list