Accuracy of Reverse Transcriptase-Derived Sequences

Pamela Norton pnorton at
Tue Nov 26 17:54:54 EST 1996

In article <01bbd8bb$dd3a73a0$cbb8c180 at>,
"Ken Soderstrom" <soderstk at> wrote:

> Dear people who know more about polymerases than I:
> I've read quite a bit regarding concern over Taq vs Vent & Pfu error rates
> in PCR.  I don't recall running-across similar preoccupation over reverse
> transcriptase error rates, and the effect that these errors have on the
> accuracy of cDNA libraries.  
> Is it acceptable to report sequence information from RT-derived cDNA
> clones?  Is it necessary to obtain a genomic clone before attempting an
> "acceptable" characterization of a gene product?  Are errors associated
> with RT "swept under the rug".
> I'm very curious to learn about what you all think.
> Thanks,
> -- 
> Ken Soderstrom
> soderstk at


     You raise a good issue. As with clones generated by PCR, the best way
to ensure the authenticity of cDNA clones is to sequence the same region
from 2 or more independent isolates. In three of three cDNA sequencing
projects, each involving several kb, discrepancies between clones were
discovered. In two instances, single clones had frame shifts that disrupted
the reading frame; I assume these were reverse transcriptase generated
mistakes. In the third case, a single, silent base difference was detected.
Two clones had one version, two the other; these probably represent a real

     To answer your question, yes it is acceptable to publish sequence from
RT-derived clones. The error rate of RT is about 10E-4, which is probably
better than the rate of errors generated during the sequencing and analysis
However, its probably a good idea to try to sequence more the one clone if
possible, at least to spot check things like open reading frames.

     Pam Norton

Pamela A. Norton, Ph.D.          Associate Professor of Medicine
Thomas Jefferson University
Philadelphia, PA 19107           p_norton at

More information about the Methods mailing list