Dr. Duncan Clark <Duncan at nospam.demon.co.uk> wrote in message
news:nPAXG8Al+we4EAcG at genesys.demon.co.uk...
> In article <85eh2u$78b$1 at ssauraab-i-1.production.compuserve.com>, Gys de
> Jongh <GysdeJongh at compuserve.com> writes
> >Dr. Duncan Clark <Duncan at nospam.demon.co.uk> wrote in message
> >news:9erByuAEgxd4EAXA at genesys.demon.co.uk...> >>
> >> Mutation frequency for Taq is 8 x 10E-6/bp/duplication
> >> Mutation frequency for Pfu is 1.3 x 10E-6/bp/duplication
> >Should this be the number of wrongly incorporated bases after the
> >duplication of one base (?) Is the dimension than 8x10E-6base/duplication
> >> So for a 20 cycle PCR of 3000bp target i.e. 1,000,000 fold
> >> amplification, errors introduced by Taq are (8 x 10E-6) x (3000bp) x
> >> (20) duplications or 48%.
> >Can you explain this multiplication ?
> >The outcome seems to be 0.48 ; why the 48% ?
>> All right the fraction of the total with an error is 0.48 therefore in
> percentage terms, 48% of the total will have an error.
>> >If you multiply the chance for incorporating one wrong base (8x10E-6)
> >the total number of bases than this will yield the number of wrong bases.
> >However this will be 8x10E-6 x 3000bp x 2^20 (why 20 ?).
> Because 20 cycles is in theory 20 duplications but why are you doing
> 2^20 when the error rate is given per duplication, which I assumed
> covered the 2^ part?
because 20 cycles IS 20 duplications but the duplication proces is not
linear. See my original post and also below.
> Assuming the error rate is 8 x 10E-6/bp/duplication I read that as for
> every base duplicated the error rate is 8 x 10E6.
Here we agree (i think) However I think that you have to remove the first /
because the dimension of the error rate is (misincorporated)bases per
(incorporated base) or duplication. Not per base per duplication (?????).
one longh ss DNA molecule which is primed. Then I think that the error rate
of 8x10E-06 means that if the polymerase incorporates the next 1 base that
there will be incorporated 8xE-06 wrong bases. This sounds a bit silly. It
is actually the change for incorporating a wrong base. p= (by definition)
the number of considered events / total events. Or in this case
p_for_incorporating_one_wrong_base = 8xE-06 wrong incorporations / one
incorporation (duplication of one base). The number is usefull in
calculations. 8x10E-6 wrong for 1 base is the same as 1 wrong for
1/(8x10E-06)= 125000 . So the polymerase will incorporate 1 wrong base in
125000 base additions. Or is this the wrong interpretation of the error rate
See : http://www.alkami.com/methods/refpoly.htm
"Misincorporation rates for different polymerases are described in terms of
errors per nucleotide polymerized, and the rate can be greatly affected by
many parameters. Several studies have concluded that different thermostable
DNA polymerases have error rates between 2.1 x 10-4 to 1.6 x 10-6 errors per
nucleotide per extension.
or : Nucleic Acids Res 1991 Aug 11;19(15):4193-8
Fidelity of Thermococcus litoralis DNA polymerase (Vent) in PCR determined
by denaturing gradient gel electrophoresis.
Cariello NF, Swenberg JA, Skopek TR
University of North Carolina, Pathology Department, Chapel Hill 27599.
DNA synthesis fidelities of two thermostable DNA polymerases, Thermus
aquaticus (Taq) and Thermococcus litoralis (Tli, also known as Vent), and a
non-thermostable enzyme, a modified T7 DNA polymerase (Sequenase), were
determined by analyzing polymerase chain reaction (PCR) products using
denaturing gradient gel electrophoresis (DGGE). The error rates were 4.4,
8.9, and 2.4 x 10(-5) errors/bp for modified T7, Taq, and Tli
PCR Methods Appl 1991 Aug;1(1):63-9 Related Articles, Books, LinkOut
Optimization of the polymerase chain reaction with regard to fidelity:
modified T7, Taq, and vent DNA polymerases.
Ling LL, Keohavong P, Dias C, Thilly WG
Center for Environmental Health Sciences, Whitaker College of Health
Sciences and Technology, Massachusetts Institute of Technology, Cambridge
The fidelity of DNA polymerases used in the polymerase chain reaction (PCR)
can be influenced by many factors in the reaction mixture. To maximize the
fidelity of DNA polymerases in the PCR, pH, concentrations of
deoxynucleoside triphosphates, and magnesium ion were varied. Denaturing
gradient gel electrophoresis was used to separate the polymerase-induced
mutants from wild-type DNA sequences. Thermolabile modified T7 DNA
polymerase, thermostable Taq, and Vent DNA polymerases were studied.
Fidelity of all three DNA polymerases was sensitive to concentrations of
deoxynucleoside triphosphates, magnesium ion, and pH. Within conditions that
permitted efficient amplification, optimization with regard to these three
factors yielded an average error rate in error/base pair incorporated of 7.2
x 10(-5) for Taq, 4.5 x 10(-5) for Vent, and 4.4 x 10(-5) for modified T7
(Sequenase) DNA polymerases.
In the above articles the error rate is the number of wrong bases
incorporated in the duplication of 1 base.
errors per nucleotide polymerized
errors per nucleotide per extension.
average error rate in error/base pair incorporated
I looked up your stratagene
reference and saw that they use the same formula for calculating the number
of wrong PCR products. I don't understand it , sorry. I found no further
explanation of their method.
>So if you do one
> duplication of 3000bp then the fraction with errors will be 3000 x 8 x
> 10E-6 or 0.024. Therefore x 100 to get to % =2.4%. 20 duplications then
> gives 48%.
Lets say for simplicity that we have 1 ss DNA molecule of 3000base. Than if
the error rate is 8x10E-06 there will be 3000x8x10E-06=0.024 bases wrong .
This is a number of bases , I think , and not a fraction or a percentage. It
means that in 1 ds DNA molecule of 3000 bp there will be 0.024 bases wrong.
Or in 1/0.024=42 ds DNAmolecule of 3000bp there will be 1 base wrong. I
don't think that this is a usefull number (see my original post) Because any
ds DNA molecule with more than 0 errors is just wrong ; so also the ones
with 1 , 2 but also 3, 4 5, etc errors. In my original post I propose to
calculate the p that 1 ss DNA molecule is made ds by the polymerase
completely free from errors. So if the error rate = p_wrong than the chance
for incorporating the right base is (1-p_wrong) In our case (1- 8xE10-6) If
we have 1 ss DNA molecule of 3000bases than :
the first base must be incorporated right p=(1- 8xE10-06)
AND the second base must be incorporated right p=(1- 8xE10-06)
AND the third base must be incorporated right p=(1- 8xE10-06)
etc upto the last base : number 30000. Than because of all the AND's the p
for that proces will be (1 - 8x10e-06)^3000 . They must all be multiplied ,
hence the power 3000. We now know the error rate for the construction of 1
ds DNA molecule of 3000bp. If we have the number of right molecules we can
than calculate the expected number of right molecules after 1 cycle. If we
want the number after 20 cycles than the proces must be repeated not
multiplied by 20 (duplication is not a linear proces ; see my original post
> > If you want to
> >express this as a fraction than you will ofcourse get the original
> >back (?)
>> This was done from memory from an article I think in Stratagene's
> Strategies vol 12 no. 1 (which is buried in the lab somewhere) but I am
> pretty sure I got it right from their article. I gave up maths a long
> time ago so leave it those who are better qualified to comment further.
>> Finally even if I'm out by a factor of 2 the original advice still
> holds. Don't use Taq if you want a finished product with few errors!
I can only agree here. More duplications or more error in 1 duplication will
yield less completely right product molecules. Still the magnitude of the
error is not what I expected. Why did I never notice that about 20% of a PCR
product is wrong ? I did SSCP of pcr products (genomic DNA as template)
after 35 cycles and found sharp discrete bands or shifts for point
mutations. If 20% is wrong than I would expect smears ??? We also do PCR
directly on colonies. The PCR products are cycle sequenced and analysed by
an ABI-310. I never saw 20% noise under the peaks in the electropherograms