Construct and compare with random network

[Shawane-Wang]

Hi. Is there any method to construct classic random network or generalized random graph in this package? And could we further compare our network with the constructed random network to characterize our network as presenting non-random interactions of microbiome? Looking forward to your reply.

[ChiLiubio]

Do you mean the random network like this (https://rpubs.com/lgadar/generate-graphs) or constructed based on the community with a null model or from a paper? I have not created a functin to directly generate it, but it is relatively easy to do. I will show an example once I am sure the random network you want to do. Comparing networks can be performed with meconetcomp package (https://github.com/ChiLiubio/meconetcomp).

[Shawane-Wang]

I mean constructing a random network using parameters from our empirical network (like count of nodes, edges and so on). Here is an example that might be vivid (https://mp.weixin.qq.com/s?__biz=MzIxNzc1Mzk3NQ==&mid=2247484753&idx=1&sn=8b00273dfee2478478d43fbbc8da7bf3&chksm=97f5b549a0823c5f673d77006ace7641cbb93539537483e6fb15243b37187498573482f3b787&token=1817505529&lang=zh_CN&scene=21#wechat_redirect).

[ChiLiubio]

Thanks. I updated the package in github and add a function random_network. Please reinstall it from github and have a try. Here is my test.

library(microeco)
data(dataset)
t1 <- trans_network$new(dataset = dataset, cor_method = "bray", filter_thres = 0.003)
t1$cal_network(COR_p_thres = 0, COR_cut = 0.3)
test <- t1$random_network(runs = 100)
test

[Shawane-Wang]

Thanks! Could we further get results like clustering coefficient, average path length rand and other network parameters according to every run of the random network, which may help us to generate the histograms of these parameters comparing with those from our empirical network? Maybe pictures as following?

[ChiLiubio]

Yes. I forgot to tell you the parameter output_sim can control whether output each result of simulation. When output_sim = TRUE, the columns from five to the last are the simulations.

library(microeco)
data(dataset)
t1 <- trans_network$new(dataset = dataset, cor_method = "bray", filter_thres = 0.003)
t1$cal_network(COR_p_thres = 0, COR_cut = 0.3)
test <- t1$random_network(runs = 100, output_sim = TRUE)
test

Then you can extract this part of results for the customized visualization.

[Shawane-Wang]

In my own dataset, I called t1 as tmp, but it returns as follows. Is it possible to be more flexible when defining our object?

I guess the error may comes from here.

[ChiLiubio]

Sorry. It is a bug. I have fixed it. Please reinstall the package from github and try again. Thanks.

[Shawane-Wang]

It works! Thank you for your kind help!