Can anyone tell my if my calculations and assumptions make sense. My gene of interest got two isoforms and I want to found out how the proportions between those isoforms change in various conditions. I got only two Taqman gene expression assays - one specific for the short isoform and the other amplifying both isoforms. If I assume that there is no bias in the amplification of a particular template (short and long isoform) I can quantify it like that:
Lets say the Ct values are
reference gene 17
The isoform one and two - 22 so delta is 5
Isoform one - 26 so delta is 9
deltadelta is then 4 2E4 is 16 so ..... isoform one +isoform two = 16*isoform two
isoform one = 15*isoform two
This is of course ideal situation were there is no bias in amplification for particular isoform and the efficiency is 100% for all the reactions. Could anyone please tell me if such a quantification makes sense and how to estimate or calculate the error for such calculation? The problem is that one of my isoforms is rare and the Ct is around 30 so probably the error in this kind of calculation is even bigger... The proportions between the isoforms in my experiment are very big when i calculate it like that. Does it make any sense or simplification of calculation creates too much errors to present this kind of results?
Regards
Tomasz
Your maths mostly holds up (if you're strictly comparing two different PCRs from the same sample, you don't really even need a reference gene), so 22 vs 26 is a 2^4 fold (i.e. 16 fold) difference, so 15:1 would indeed be the interpretation.
And you are correct in accepting that efficiency differences can affect this (can affect it substantially, in all honesty).
If you wanted to be more thorough, you could prepare purified PCR products of your two primer sets and run a standard curve of those (diluted from millions of molecules all the way down to stochastic levels just to be sure). This will give you your efficiencies. To be super safe, then do the SAME dilution curve in the presence of fixed concentrations of your test samples: the curves should flatten off at the point where your standard curve concentration matches that of the test samples, so this should give you a pretty good idea as to whether your numbers as estimated from the above calculations are in the right ballpark. If one target is giving numbers in the 30s, do additional replicates to minimise stochastic variation (like, 5 or 6 replicates rather than 3).
Finally, however: are you sure expressing the two as ratios is appropriate under these conditions anyway? If you have two forms and one is present at vanishingly low levels, this isoform is not significantly affecting the level of the other.
Imagine isoform 1+2: 10000 transcripts. Isoform 2: 10 transcripts.
Now in your test samples: isoform 1+2: 10000 transcripts. Isoform 2: 20 transcripts.
Does expressing this as 10:9990 and 20:9980 actually add any extra information? Basically your levels of transcript 1 (as extrapolated by subtracting level of isoform 2 from the combined total) do not change under these condtions (9990 is the same number as 9980, within biological accuracy), while your levels of isoform 2 double (10 to 20).
I would only start worrying about the contributions of isoform 2 if the levels are actually appreciable (like, maybe 5-10% of the total signal). If they're vanishingly low, then just treat your "total" PCR as being a measure of isoform 1, and your isoform 2 specific PCR as being a measure of isoform 2.
Thank you for the fast and very thought provoking answer - I am much more encouraged to continue my calculations.
What happens if you do a limiting doubling dilution of your samples and do PCR amplification with the 2 Taqman probes until both signals are gone? Presumably you will reach the point of no signal with one probe before the other? The difference in the fold dilution where loss of signal occurs between the 2 will inform you of their relative amts.
Just realized this is an ancient post and you have probably dealt with the issue by now ...
- Badges
- Science method