how to analysis technical replicates data?

Question

[Yizhao]

Hi all,

I've got protein MS data that contains 3 technical replicates. For all I know, technical replicates can examine the repeatability of technology. So, I will focus on those overlapped proteins between all 3 replicates. After that, I want to conduct differential expression analysis. Here comes the question. How do I do with the data from 3 replicates? Average 3 protein density values of 3 replicates as the protein's density? Is that OK? OR some other good ideas?

Any advice will be appreciated.

THX in advance.

best,
Yizhao

Answer 1

[Asaf]

I think you should take the sum of all technical replicates.

[Yizhao]

Thx, and now, I have the same idea with you.

Answer 2

[Woa]

You can use technical replicates to measure the variance(equivalent to noise ) of your data, and above the observed variance, you consider everything as signal. Often fold-change cut-off to ascertain differential expression is determined this way.

See this paper and related papers from the same group.

[Asaf]

But then you won't take into account biological noise.

[Yizhao]

Yes, I can do nothing to avoid the bias caused by sampling, or biological noise. But, actually, my experimental samples were sampling through a time line. At present, I have no idea about the analysis with time series data. Is there a practical strategies, except that comparing pairwise time-point by fold-change?

[Yizhao]

Yes. Fold-change cutoff was my choice in RNA-seq analysis. Thx a lot. Yizhao.