Hi,
I'm new to qPCR, so please forgive me if my question may seem stupid. I am using a PikoReal machine to quantify viral load (DNA) with the help of a probe. So I'm testing viral DNA in bird DNA samples and am always adding a positive, as well as a no template control to each plate.
My problem is that a lot of my samples are weakly positive, so with Cq>35, and the automatic baseline calculated by the PikoReal software is therefore often too low (please see attached PDF for examples). This results in dodgy duplicate values (e.g. 24 in one well, NaN in the other well for the same sample), because the software is actually measuring background noise in some wells because the baseline is set too low.
I am not sure how to adjust the baseline in a way that
a) the results are not "fake" (e.g. by cutting out weak positives by lifting the baseline too much), but as close to the truth as possible
b) the results are comparable across many plates, as I have 100s of samples.
Any help would be much appreciated :)
Hello Johanne,
what kind of cycler do you use? Is there a way to remove the background noise (eg by using an outlier removal)? Do you use standards with a defined concentration ? If so, you could dilute them further to establish a limit of detection: Enclosed you will find my suggestion for a manual baseline (red).
best regards
Dirk
Have you run a bird DNA only control? Perhaps try other non-bird genomic DNA controls and see what you get with them, maybe if you're probe has a higher than ideal GC content it may be binding non-specifically more than you'd hoped! I agree with Dirk regarding a virus standard DNA; I would run a standard curve and see what Ct values you obtain for a minimum meaningful level of virus DNA.
Another way might be to set the baseline to a consistent control value. The trace that comes up at around 15-20 cycles has a Ct value of 17.4 in the first trace and 18.2 the second. Presuming the noise on all your runs is of the same magnitude you could for instance set the threshold such that the Ct on the control is always exactly 18.
Ok, reading your question and looking at your data, I would honestly say that your "weak positives" are in fact noise, and should absolutely be dropped from your dataset. As a rule of thumb, a single template molecule should produce a Cq value of around 35, with 2 molecules thus being ~34, four being ~33, eight ~32 etc.
Of course, with template levels this low, you are well into stochastic partitioning territory, so replicate wells of a low [template] solution might end up with 2, 6 and 1 template per well, respectively (giving you high Cq variation between replicates). It also means that even a tiny fractional contamination of something like your primer stock is a real risk.
Also, the fewer template molecules you have, the greater the chance of primer dimer formation giving you a false amplification curve.
This is why data giving Cq values above 30 should always be treated with caution, and why you should always perform a post-run melt curve (assuming this is SYBR green qPCR -you don't specify).
For your samples giving Cq values greater than 35: ditch them.
For samples giving Cq values of 30-35, repeat, but with a greater number of replicates (stochastic partitioning effects can be circumvented by greater replicate number: essentially giving you a better average). Also, include the controls as suggested by Marcus: non-avian DNA, or guaranteed viral-free avian DNA.
Another thing to consider would be to change the amount of DNA per well: if you load twice as much, does the Cq value reduce by ~1? If you load half as much, does it increase by ~1?
If so, you can at least ascribe the measured value as being relevant to your sample (as opposed to primer contamination etc). It doesn't rule out sample contamination, but still.
Thanks so much everyone for your help and suggestions!! I am using a PikoReal 96 well system. Yes, I can manually remove "noisy" samples if necessary. It's not a SYBRgreen method, so I can't get a melt curve.
I dilute all my samples (viral + avian + other DNA extracted from blood sample) to 200 ng/ul and then add 2ul of this into the well.
I run every plate with a positive control, which consists of DNA pooled from several strongly virus positive birds, and with a no template control (buffer).
I've run serial dilutions with some of the positive samples that I've then used for the pooled positive control. They looked good, so loading half the amount resulted in a Cq decrease by 1, etc.
A previous PhD student used the same assay, but with an older PCR machine. He found that the cutoff for him was a Cq of 38. Below this (e.g. Cq 35), results were highly repeatable. But I guess this can be different with a different machine, so my cutoff may be 35?
Also, to make results comparable across plates, I guess I can normalize the results by calculating deltaCq with the positive control on each plate?
Dear Johanne,
if you are using dye labeled probes instead of SybrGreen you can very well perfom melting curve analysis. That´s the good thing about dye labeled probes. Try it.
Best regards
Dirk
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- DNA