Hello everyone, I'm trying to use the CoExpNetViz, Cytoscape plugin to individuate significant expression correlations (in a co-expression network) between a reduced number of genes of interest (9) and a larger number of them in a dataset.
I ran into a weird behaviour... It seems that, if the number of genes to investigate is relatively high (>500), the results are not reproducible. I get a slightly different network plot every time, and, for certain datasets, no plot (only a blank screen).
For instance, out of 5 runs, I can obtain some, or all, slightly different correlations. The greater the number of genes to investigate, the more pronounced the differences among the attempts. I also suspect the result is truncated, because the significant correlations aren't as many as I would expect, especially for larger datasets. The problem only occured with a dataset of at least 500 genes, approximately. I ran this kind of test on a smaller one and the results didn't vary.
Here is an example: on a dataset of 1000 genes, out of 5 runs, I get 3 different results, individuating a slight different number of target genes for each case. Pearson coefficients don't vary for the mutual genes.
https://i.ibb.co/vvn32mg/Test1.png
https://i.ibb.co/pZMxvkX/Test2.png
https://i.ibb.co/1bPK5kW/Test3.png
At first, I believed the problem could be memory-related, so I tried, as suggested in Cytoscape 3.9.1 manual, to allocate more space, but it made no difference. Also uninstalling and reinstalling the plugin didn't change anything, nor using a computer with a higher RAM (I have 64 GBs right now).
Has this happened to anyone? Thanks in advance.
Greetings,
I would suggest reaching out to the authors of the app. Be prepared to send them some examples networks so they can debug the problem.
-- scooter