error free proteins

Roger Wiegand rcwieg at
Tue Oct 25 08:59:47 EST 1994

In article <40ABAC2E01B938D9 at>, varyc.mmcri at OFFICE.MMC.ORG
(Varyc) wrote:

> Dear netters,
> I enjoyed reading B. Foley's summary 'calculations' in methods. I am curious
> as to readers' opinions on the most error-free polymerase(s) to employ for
> RT-PCR. Specifically for the  purpose of  1. cDNA synthesis from a mRNA and
> 2. pcr amplification to end up with an error free clone for protein
expression. As
> both the RT and the Taq pol's lack 3' editing error can creep in at any stage.
> My concern stems from the observation of too many sequence changes for
> comfort when scanning clones for restriction sites or other more direct
> measures on sequence integrity.
> Cal Vary
> Maine Medical Center Research Inst.

The only wat to get (close to) error free proteins is to sequence multiple
(completely) independent isolates of your cDNA (or sequence the genomic
clone from the same source), then completely and carefully sequence your
cloned PCR product. There is to my knowledge no low error rate RT
available. The error rate in PCR can be lowered by limiting the number of
amplification cycles, using primers that work well, and by using a
polymerase with a proofreading function. I routinely run multiple PCR
reactions side-by-side and then pool the products for cloning--that way
when I sequence a bunch of them I increase the likelihood that either the
pool contains completely correct clones, or that I will pick clones with
different errors that can be stitched into a correct sequence by cut and

Even being careful with the PCR conditions I have seen errors resulting
from all of the currently available polymerases--and on two occasions even
though the gene sequence was correct the primer had been made incorrectly
by the (former) commercial supplier. I've found that about half the
"errors" we've seen have been in the RT or PCR steps and about half as
incorrect Genbank entries.

If you want to be sure, you have to sequence your clone, both strands, all
the way through. Sorry.


mailto::rcwieg at

More information about the Methods mailing list