Dear Andrea Magnavacca

Given that you have two culture conditions - |Normoxia| and |Hypoxia| - it translates to having two classification variables (or 2 test groups). This would mean One-Way ANOVA is not appropriate. You will need to use a T-TEST.

Whatever the convention you decide to use to compute ∆Ct of your genes depends on you; some people use ∆Ct = Ct goi - Ct housekeeping ……..(1)

whereas some others use ∆Ct = Ct housekeeping - Ct goi…………(2)

The interpretation of ∆∆Ct is dependent on the method you choose in the computation of ∆Ct from Eqns. (1) or (2). With Eqn. (1) you have to interpret/speak inversely when you compute ∆∆Ct.

With Eqn. (2), when you compute ∆∆Ct, then its interpretation is direct. Positive values indicate up-regulated genes, and negative values indicate down-regulated genes. Kindly see other similar posts on Researchgate as there are a number of posts which address your question.

You can test for statistical significance of your ∆Ct values using Microsoft Excel. The following steps works to activate Statistical Analysis ToolPak on versions 2010 and 2013 of Excel. The >> indicates “click”

Start Excel >> File >> Option >> Add-Ins >> Go (at bottom of page for "manage Excel Add-Ins") >> Check “Analysis ToolPak” >> OK

This should activate the Excel’s Data Analysis tool. If you use another version, kindly check the Microsoft Support Web for guide on activating the Toolpak for your version.

I believe you can also feed the calculated ∆Ct into Prism and test for difference between the two groups using the T-TEST procedure of the software.

I hope this helps.

Kind regards,

Dami.

I would use limma. Although it was designed for micorarrays (and is recently used for NGS data) it works equally well with dCt value form real-time PCR.

https://bioconductor.org/packages/release/bioc/html/limma.html

It could be Limma, or Anova.

Thanks for your answers, but we have neither R nor Limma skills. Is it possible to analyse this kind of data using Prism? Even only to asses significance in the differences of expression of each gene between control and hypoxia.

Dear Andrea Magnavacca

Given that you have two culture conditions - |Normoxia| and |Hypoxia| - it translates to having two classification variables (or 2 test groups). This would mean One-Way ANOVA is not appropriate. You will need to use a T-TEST.

Whatever the convention you decide to use to compute ∆Ct of your genes depends on you; some people use ∆Ct = Ct goi - Ct housekeeping ……..(1)

whereas some others use ∆Ct = Ct housekeeping - Ct goi…………(2)

The interpretation of ∆∆Ct is dependent on the method you choose in the computation of ∆Ct from Eqns. (1) or (2). With Eqn. (1) you have to interpret/speak inversely when you compute ∆∆Ct.

With Eqn. (2), when you compute ∆∆Ct, then its interpretation is direct. Positive values indicate up-regulated genes, and negative values indicate down-regulated genes. Kindly see other similar posts on Researchgate as there are a number of posts which address your question.

You can test for statistical significance of your ∆Ct values using Microsoft Excel. The following steps works to activate Statistical Analysis ToolPak on versions 2010 and 2013 of Excel. The >> indicates “click”

Start Excel >> File >> Option >> Add-Ins >> Go (at bottom of page for "manage Excel Add-Ins") >> Check “Analysis ToolPak” >> OK

This should activate the Excel’s Data Analysis tool. If you use another version, kindly check the Microsoft Support Web for guide on activating the Toolpak for your version.

I believe you can also feed the calculated ∆Ct into Prism and test for difference between the two groups using the T-TEST procedure of the software.

I hope this helps.

Kind regards,

Dami.

Agree with Damilola E. Akinyemi regarding the ∆Ct calculations and interpretations.

However, a one-way ANOVA with two groups is equivalent to a t-test.

The t-tests are simpler to perform, in Prism as well as in Excel (=T.TEST({dCt values of Normoxia};{dCt-values of Hypoxia}; 2; 2)). No need to use any Analysis Tool Pack.

The drawback (as compared to limma) is that the ANOVAs or t-tests don't use the (variance) information from all the genes. You would do one test per gene, using only the information of the respective gene.

I feel dr Jochen has rightly mentioned.

1) You have 48 statistical tests using ANOVA (pairwise comparisons using t-test). This gives you a fair chance of finding at least one "significant" difference even if there is no significant difference between the two treatments. You can either figure out how many significant outcomes would be above random chance, or you can reduce your significance threshold by the number of tests (0.05/48).

2) You might consider something like discriminant analysis as a way to look at all 48 genes and estimate their combined power to discriminate between your two treatments. However, discriminant analysis will not work with 48 variables and only four replicates. At a minimum, you need more replicates than you have variables.

3) You need to do a good analysis based on published papers with similar data. If PRISM cannot do a required analysis, then you need to get better software or you need to partner with someone who can do the analysis. There are many programs that will work: R, SAS, Statistica, JMP, Minitab, SPSS, and many others. Here is a list of free software, though there is no guarantee that all of these will do all of the analyses that you need: https://www.predictiveanalyticstoday.com/top-free-statistical-software/