search

Vivianstats / scLink

Inferring gene co-expression networks from single cell gene expression data
https://www.sciencedirect.com/science/article/pii/S1672022921001455
24 stars 7 forks source link

Some questions about simulation process #8

Closed HelloWorldLTY closed 1 year ago

HelloWorldLTY commented 2 years ago

Hi, I wonder how to generate a simulation process based on the description you have in the paper. In the method part, you state that you first generate a correlation matrix and then transfer it into an adjacency matrix based on uniform distribution. Therefore, why this adjacency matrix is not the final edge matrix of the graph? Since it seems that you mentioned the distribution of edges follows some rules: image Do the statements conflict? Thanks.

Vivianstats commented 2 years ago

Hi, just to clarify, we first generated a connectivity matrix and then generated a partial correlation matrix based on the connectivity matrix.

HelloWorldLTY commented 2 years ago

Hi, many thanks for your reply!

HelloWorldLTY commented 2 years ago

Hi, I meet a new problem when I intend to simulate the hub datasets: image I try to control the ncluster>degree.hubs, but still, receive this error. Could you please help me solve it? Thanks a lot.

Vivianstats commented 2 years ago

This is an internal error from the ldstatsHD package, so I would suggest that you reach out to the developer of this package for this question.

HelloWorldLTY commented 1 year ago

Thanks a lot!. Btw, I also found a warning when I normalized data for sume datasets then run the sclink: Warning: the standard deviation is zeroConstructing gene networks ...

If I encounter this error, even if I try to infer co-expression relation from the same tissues, the results are different. Is it the problem of my own dataset? Thanks a lot.