Across-time co expression analysis

Question

[HERMAN]

Hi BioStars !

I am currently trying to analyse a piece of data from mouses and don't know how to do it. The experimental design is bulk RNA sequencing of 10 cerebellum at 10 development stages each. I have now a data frame, with row genes and column timepoints filled with mean expression values . I'd like to know if you guys knew some packages, or at least some key-words to dig in myself, to let me do a co-expression analysis over the time points ( e.g which genes are up regulated together over the 10 stages ) ?

Thank you for reading and for your time,

Cheers, Simon

Answer 1

[LChart]

For developmental timecourse analysis, WGCNA has been applied successfully in the past (https://www.sciencedirect.com/science/article/pii/S0092867413013494, https://www.sciencedirect.com/science/article/pii/S0896627314004486); but note that the expression is not aggregated to mean values ahead of time, but rather sample-level data is used directly. There will be an impact if different timepoints have different numbers of replicates.

For pre-aggregated data like this, when I'm seeking to explore visual patterns across the timepoints, I will typically Z-score the expressions across timepoints (within genes) and run a gaussian mixture model using mgcv mclust, using BIC plots to determine the number of components. There's also a deep literature on model-based network analysis.

[HERMAN]

Thank you very much for your response, I'll give it a try !