Hello,
My colleague and i have a linearized double stranded DNA fragment that we want to amplify through ddPCR. The problem is we're only getting a %50 recovery with it. We've double checked and we know our reagents (primes probes and master mixes) are fine since they work with other similar assays we've used. The machine too has been calibrated and checked. He made sure to measure the DNA as well to make sure there isn't any issue with the quality or reported concentration. The only other issue was an potential freeze thaw damage (although that seems highly unlikely). To counter this he's frozen them down in aliquots to avoid any potential problems. Does anybody have any similar experience or know what other potential issues could be causing this low recovery rate? Thank you!
Hi Matthew,
let me understand better your problem.. what do you mean by recovery? The estimate of number of molecules is halved? Did you get enough droplets? Have you tried different concentrations/dilution to see if the reduction between expected and observed counts is maintained the same?
I must sadly and repeatedly say that PCR is not a quantitative method at all.
DNA polymerase seems to have low affinities to substrates; monomer DNA polymerase enzyme has Km values c.a. 50 μM to dNTPs, but dimer human serum biotinidase has Km 0.48 μM to biocytin (please see files; BIN Mr 110,000 and The Fascio Effect).
Droplet digital PCR (ddPCR) is also not quantitative, although it does not use directly fluorimeter and/or photometer such as real time PCR (RT-PCR). ddPCR uses parametric statistics (numbers of positive wells and numbers of total wells are treated by the Poisson distribution (parametric distribution)). Then, I must sadly and repeatedly say that nature does not obey the parametric distributions such as Normal distribution and/or Poisson distribution. Nature obeys Non-parametric statistics as shown in the file "Dr. Terentyeva Urine BIN".
Therefore, the study about the gene expression should be performed by our quantitative PDMD (Protein-Direct-Microsequencing-Deciphering) method (please see files; HepG2 Fucoidan and JMBT Alopecia).
Hi Matthew,
Are you using fragmented or unfragmented DNA?
If it is unfragmented, it is possible that your DNA molecule is too long (large) to fit in droplets in a evenly distributed manner.
If it is fragmented, especially random fragmentation (e.g. cfDNA), the observed count is usually lower than expected value. The longer the amplicon by the primers, the lower the counted value. This is due to actual availability of the target sequence among your input DNA (if the fragmentation occurs at target sequence, no amplification can be done). For your interest, cfDNA has mean length of 160bp only.
For the statistic calculation, I wonder how Mr. Hayakawa has experienced this. Distribution is calculated as Nonparametric usually due to low sample size and observe outliners.
For ddPCR (assume you are using Bio-rad ddPCR like me), the sample size is large n>10,000 /test, and skewed. Also mean is the answer we need (nonparametric is better in finding median). I would say the Poisson calculation is good enough. I have obtained good correlation between 0->80,000 target copies/well.
Hope it helps
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- DNA