search

petitmingchang / TO-GCN

Pipeline of time-ordered gene coexpression network (TO-GCN) construction from three-dimensional (gene expression, condition, and time) data
39 stars 22 forks source link

The Cutoff positive coexpression for Count is used 1 is 0.00 #16

Open shengyuedai opened 9 months ago

shengyuedai commented 9 months ago

Hello, your program is very useful for our experiment. When calculating the first step Cutoff, the Cutoff positive coexpression for Count is used 1 is 0.00. Is this normal? How should I correct it if it's not right? CUToff

jordan841220 commented 9 months ago

Same issue here when I input VST values. I'd like to know the answer as well.

petitmingchang commented 9 months ago

@shengyuedai Can you provide your input data for testing? thanks!

shengyuedai commented 9 months ago

Thank you very much for your reply. Here is my data. I have two treatments, each with three time points. My main focus is to identify downstream genes that transcription factors may act on. Thank you very much for your help! NaCL 0h、12h、24h；NahCO3 0h、12h、24h.zip

petitmingchang commented 9 months ago

@shengyuedai and @JordanCTLin The TO-GCN method highly relies on the PCC (Pearson Correlation Coefficient) value to construct the network, so we suggest that "at least 5 time points" were needed for each condition. if data with only three time points, PCC values for most gene pairs will greater than 0.99 or less than -0.99. That's why we did not get good cutoff values.

shengyuedai commented 9 months ago

Thank you very much for your reply. I currently only have three time points. What is a good way to obtain a reliable PPC value? I really want to use TO-GCN for data analysis

petitmingchang commented 9 months ago

@shengyuedai If you have three time points only, the expression pattern for each gene would be too simple to construct a meaningful TO-GCN. I will suggest using clustering method to group genes with similar pattern for each condition. Then you can compare genes in similar pattern between two conditions.

shengyuedai commented 9 months ago

Thank you very much for your suggestion. If I continue to do this with a Cutoff of 0.00, can I refer to the results?

petitmingchang commented 8 months ago

@shengyuedai For sure you can use extreme Cutoff value to get the result. However three time points are too few to get a reliable interpretation.