Amplifying from pcr products - (Aug/20/2014 )
Is it true that amplifying from pcr products that you already have gives you many problems rather than doing it from the genome? it seems counterintuitive to me.
It depends on what you are trying to do and how you are doing it - nested PCR can be very efficient and works extremely well, but if you were trying to clone from a PCR on top of another PCR, then you are likely to introduce a bunch of sequence errors (depends on polymerase etc.).
I see, it just seems to me that besides methylation, the pcr product is just another DNA molecule, i do not see why the argument of introducing mutations is so strong in here, since that should be true when doing it from the genome. I'm not disagreeing i just want to understand the reasons and see if anyone knows them. Note that i mean a pcr product that has been cleaned and diluted to serve as template.
With regards to the mutations - yes it is a problem, Taq has a mutation rate of about 1 base in 10,000... now that doesn't mean that you would need to have a product 10000 bases long before you would see an error, it means for every 10000 bases amplified you will see an error (on average)... here's the math:
100 bp PCR product - assume a single copy to start with. At the second cycle you now have 200 bp copied, third 400, 800, 1600, 3200, 6400, 12000 (note you can now assume one error), the next cycle you have 24,0000 (so that's 3 errors 12000 + 24000), and so on in logrithmic progression, by the end of 30 cycles you will have a lot of errors, and that's just for 100 bp - imagine you were working with 1000 bp!
You then take those products and amplify them again....
For some frame of reference - you might use 10 ng of starting PCR product for your second round amplification - for a 100 bp product this is about 9.9 x10^9 molecules - imagine the number of errors in that!
Yeah, introduction of errors in polymerases is a memoryless process. Thanks for the explanation, that makes sense.
Reamplifying and Taq errors can be a problem, indeed, but in my experience (also with some reamplification or amplification with high number of cycles) for example for sequencing purposes it's still very usable.
I've been sequencing PCR product for years and never had a problem with a "false" mutation in my sequencing reads. It's true that after nested PCR you have many molecules with single-base errors in them, but you also need to consider the overal amount of them. They are mutated randomly so it's possible that for every single base pair in 300bp product there is a mutation in one or even hundreds of the bilions of copies, but overal the majority of the amplicons carry the right base at every position.
Since sequencing doesn't detect variants below 10% of overal, it is very unlikely you will get amplicon that has 10 % or more of incorrect base. It can only happens if the mutation occurs in very early cycle (since every mutation is them copied from it's template and multiplied). For that reason, if I find a congenital variation in my patient sample I always check by second PCR (and ideally by second blood sample).In the second PCR the odds of the random mutation rate are competely different, so if it's an artefact, you won't see it in second PCR.
So, for sequencing, if you have in mind that rarely you can get "false" data, you can use it.
But, as mentioned, it's not suitable at all for cloning, I used the same polymerase for cloning only once (reamplified), I needed just a PCR control - amplicon cloned to the plasmid. I needed just the primer sequences to be intact and didn't care about the rest of sequence, that's why I didn't go with proofreading polymerase.
I had a 400bp or so product and sequenced it from the miniprep. Of four clones every single one of them had a mutation within the sequence, one even more. I could use the clone that didn't have mutation in primer sequence, but that shows you how actually precise classic polymerase is.
Some other aspects (especially the topic mentioned by homebrew) of that problem can be found in this interesting thread: click me
Yes of course, for finding mutation in a very small portions of original template (or for bisulfite sequencing for methylation pattern) it's better to sequence clones. But for a casual sequencing for congenital mutation this is just overkill, you are testing for example about 10 genes, each has several exons.. that would be a tremendous amount of work for cloning them all.
Cloning product is fine for some uses, but needless for other.
Finding rare mutations from blood DNA may be actually very important for diagnosis of diseases, but Sanger sequencing seems to be limited in this way, as states new paper about deep NGS resequencing.