If i get the right idea of your question, I say you have 20 observations (caution: if all 20 genes have exactly same structure and your medium are exactly same for all of them. otherwise you should consider them as subjects) with e.g. 5 time intervals, you can use one-way anova with 5 groups.

But if your data considered to has subjects, you can use one-way Repeated measure. In repeated measures be careful of sphericity assumption (result of mauchly's test), if it violated, you cant rely on the results, but it depends.

The kind of analysis depends on *your* research question. This question is the *start*, from which you plan the whole study (what data to measure, how to measure, what to sample, how to sample, ...). Already before you have the data it is (or should be!) thus clear how the data have to be analyzed to best help you answering your research question. Statistics is nothing that comes after the experiment. Statistics provide tools to quantify how much you can learn from data and is thus deeply involved already in the planning phase. I would suggest that start learning statistics (what is "knowledge", what is "information", what is "learning", what is "probability", what is "likelihood", what is a statistical model, what is experimental design, ...??) and then start over doing the science.

I will give you a more specific help for your problem when you can correctly state (i) what the p-value you are aiming to calculate exactly means and (ii) why exactly you want to calulate p-values (or perform such statistical tests) ;)