Question: microRNA network analysis
1
gravatar for lessismore
8 months ago by
lessismore490
Mexico
lessismore490 wrote:

Hey all,

i have a list of microRNA (almost 300) for which i have raw counts in control and a stress condition for 2 genotypes. I was thinking to use the absolute values (after normalization) for a network construction for each genotype in the 2 conditions (stress and control) and compare these networks in order to find conserved modules between the genotypes.

Furthermore i would like to insert in some way the data i have from a target prediction analysis. But i have no RNA-seq data for the mRNA.

Any ideas or suggestion for facing these steps? Every tip (papers, softwares, guidelines) would be much appreciated

ADD COMMENTlink modified 8 months ago by Kevin Blighe22k • written 8 months ago by lessismore490
0
gravatar for Kevin Blighe
8 months ago by
Kevin Blighe22k
University College London Cancer Institute
Kevin Blighe22k wrote:

Hey / Buenos días, it looks like you have interesting data.

My first question is why you are thinking about using absolute values? Of the network construction methods that I know, negative values are not a problem as they are based on eigenvector/eigenvalue transformations (WGCNA), or they just build a distance matrix as that starting point.

Based on your experimental set-up with with genotypes, I recommend CGBayesNets by my colleague Mike: http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003676 It's implemented in MATLAB, but it does specifically allow you to see how a network varies by things like genotype.

ADD COMMENTlink written 8 months ago by Kevin Blighe22k

Hola Kevin, thanks for your answer.

My first question is why you are thinking about using absolute values? Of the network construction methods that I know, negative values are not a problem as they are based on eigenvector/eigenvalue transformations (WGCNA), or they just build a distance matrix as that starting point.

Ive been reading that you can use either absolute values (in my case TPM) or log2 values of the fold change (Control/Stress). So you think i could make 2 trials on that?

ADD REPLYlink written 8 months ago by lessismore490

Hey, what do you mean by 2 trials?

I would be very interested in seeing how the network changes in the stress condition based on the 2 genotypes. You could build 2 different networks in order to do this, and them compare them (for example, compare community structures - 'communities' are akin to modules in WGCNA).

ADD REPLYlink written 8 months ago by Kevin Blighe22k

My question is: would you use your TPM or the log2 of the fold change control/stress?

ADD REPLYlink written 8 months ago by lessismore490
1

For the network construction, I would use the TPM values. It should be possible to then highlight any of your genes that were significant to see in which modules they group in the network plot.

ADD REPLYlink written 8 months ago by Kevin Blighe22k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1963 users visited in the last hour