Hello everyone,
I am trying to determine relative expression values for specific genes in different life-stages of my organisms (adult, larval, and microfilaria). For each of the three life-stages, I have three biological replicates and performed all PCRs in triplicate. I have two reference genes to compare with my genes of interest, but I am unsure how to calculate fold expression changes if there are no treatment groups/control groups with the ΔΔCt method since I am only assessing life-stage differences in expression. Also, with two reference genes, I am unsure at which point in the analysis I need to account for this.
Any advice would be greatly appreciated!
Hi Jennifer Nichols
If you do not have control group, then use the absolute comparison by drawing a standard curve.
For two reference genes, I do not have an experience in case of absolute comparison, but in the relative one we calculate their geometric mean.
For multiple references genes, you calculate either the arithmetic mean (for Cq values) or the geometric mean (for linearised relative quantities). This is because Cq values are approximately normally distributed, whereas linearised RQ values are lognormal.
Also, well done for using two references: this is very good practice.
For your actual data, you don't actually need to calculate absolute values (and indeed absolute values won't necessarily tell you much more than relative values + "rough eyeballing of the raw Cq" will): what you're interested in is the difference in expression between your different life stages, and this is an entirely relative comparison. You could, in essence, pick a single life stage (any life stage) as your "control" and then relate all your other life stages to that.
So, here's a basic workflow: you have one GOI and two REF for larval, microfilaria and adult, with three biological replicates in each, PCR'd in triplicate.
- Calculate the mean Cq values per gene, per sample (i.e. the average of your triplicate reactions). Look for any outliers and discard accordingly (this is why you do this in triplicate: it's not uncommon to get "22.3, 25.4, 25.3" or similar, and the approach is generally just that "assume the 22.3 is a well where something went weird, so discard")
- Calculate the per-sample mean REF Cq, which can be arithmetic, since you're in log space (where PCR data is normally distributed): so for adult #1, that's 0.5*( adult#1 REF1 Cq + adult#1 REF2 Cq). This is your "normalisation factor" (NF) for that sample. At this point it's worth looking at your NF values for your dataset as a whole and confirming that they're pretty consistent: anything with very low or very high NFs might be outliers, and any consistent deviation (i..e "larval NFs are always 3 cycles higher") is a sign your references are not good references.
- Calculate the per-sample normalized GOI expression (adult#1 GOI Cq - adult#1 NF). These are your dCt values, and...honestly, this is where you can stop. All your comparisons will be between dCt values, so you don't actually need to do any more data manipulation. You can, if you wish, invert everything (flip the sign, so 3.4 becomes -3.4) because for dCt low values indicate high expression, while "high values = high expression" is more intuitive. Just put "-dCt" on the Y axis.
- Note that it is very good practice to present qPCR data like this (i.e. in log space) because here changes both up and down (10 fold up, 10 fold down, etc) will be given equal apparent weighting in your plots.
- Additional changes: if you want to standardise your data such that one of your groups is distributed around "0", you can do that. It won't actually change any of your data, because it will be an en bloc transformation. For this (since you're still in log space), you determine the arithmetic mean of whichever group you want as your 'reference' and then just subtract that mean from all data (usually your would invert your data first, to give -dCt as above). Then just plot that. If you picked "adult" as your 'reference', then all your adult values will now cluster around 0, while larval and microfilaria; values will be either more or less than that.
If you want to share some sample data, I'm happy to work through an excel version of...well, all the above, if you like.
Since you don't have untreated group or calibrator, I would measure 2^-delta Ct instead. This will give you the expression of your target gene relative to your internal control (housekeeping gene). And the qPCR data will be presented as normalized individual data points.
The normalized individual data point (e.g., 2^-delta Ct) is identical to the 2^-deltadelta Ct; only the sample is not related to a calibrator (hence, only one dCT is present in the equation).
The 2^-Ct form of the equation may be considered the non-normalized, individual data point. The only difference between 2^-delta Ct and 2^-Ct is whether one wants to include an internal control or not. Almost always the Ct from a PCR should be compared to an internal control. Exceptions include when one wants to determine whether the internal control gene changes during the experiment.
- Badges
- Science topic