PCR fidelity

[Frank Chen]

On Thu, 21 Dec 1995, Ned Mantei wrote:

> In article <4b9cu7$i26 at thorn.cc.usm.edu>, rbateman at ocean.st.usm.edu
> (Robert Bateman) wrote:
> 
> > I have a graduate student who is using RT-PCR to amplify a large section 
> > of the coding region (75%) of a peptide processing enzyme from several 
> > mammalian species. We intend to sequence these and compare the sequences 
> > to identify conserved residues. His committee is concerned about the 
> > fidelity of Taq polymerase even though he is directly sequencing the PCR 
> > products. 
> 
> 
> If he is *directly* sequencing the PCR products, without cloning them,
> then he is reading the *average* sequence. The 0.1% or less errors at each
> position never show up. 

It may not be true.
It is probably true that errors in Cycle sequencing will not affect the 
final sequencing result of a pool of template samples.  However, the 
error in initial PCR reaction could show up IF the starting material is 
minimal.  Errors in early amplification cycles can accumulate and generate a 
pool of templates as the final PCR product.  It is just like statistical 
analysis: the average is meanless if the sample number is not big enough.

But, I agree that sequencing the PCR product directly will have much 
better chance in reviewing the real sequence.



>There would only be a problem if the polymerase
> always made a mistake at >25% or so frequency at one particular position,
> something for which there is no evidence at all. The problems with
> mistaken incorporation arise when one looks at individual clones. I don't
> think the committee understands the problem.
> 
> -- 
> Ned Mantei
> Dept. of Neurobiology, Swiss Federal Institute of Technology
> CH-8093 Zurich, Switzerland
> mantei at neuro.biol.ethz.ch   Fax: +41-1-633-1046
> 
> 
Finally, high fidelity Pol may reduce the error rates to a point.  But, 
the final answer to this kind of question is to compare your sequence to 
the genomic sequence.

Frank Chen