Context dependent polymerase errors [was Re: Sequence variations]

[Oberste]

In article <222 at sftwks.UUCP> bradbury at sftwks.UUCP (Robert Bradbury) writes:
>In article <Bz1w8o.ExA at ncifcrf.gov> oberste at fcs260c.ncifcrf.gov (Oberste) writes:
>>Since a PCR-induced base change is propagated through subsequent cycles,
>>it is important to collect and compare sequence from independent PCR
>>amplifications. It would be extremely unlikely to see the same artefact
>>in independent amplifications.
>>
>I'm not sure that I agree with this.  Polymerases will incorporate
>improper bases depending on the surrounding context.  If a
>polymerase is inclined to make a mistake once, it might make the 
>same mistake again.  

Sure, but since the error frequency is fairly low to begin with, consecutive
amplifications will _most likely_ not have errors at the same place.

>                     A much better solution IMHO is to do the PCR
>with several different polymerases.  The polymerases tend to show
>different misincorporation characteristics and it would be less
>likely that different polymerases would make the same context
>dependent errors.
>
Excellent point. 

>Another thing to consider is the error correction capabilities
>of the polymerase.  Those without 3'-5' exonuclease activites
>have an error rate about 100x those with this exonuclease
>activity.
>
Even more excellent point :-). You'll still want to look at multiple clones,
though.

-Steve

-- 
Steve Oberste                            Internet: oberste at ncifcrf.gov
LCMS, PRI, NCI-FCRDC
PO Box B                        "Never put off until tomorrow that which
Frederick, MD 21702-1201            you can do the day after tomorrow"