Open cyycyj opened 2 months ago
Let me give an update about the detailed process:
Step 1: Identify the transcription factor (TF) genes and non-TF genes.
Step 2: Perform DEG analysis between 5 stages with DESeq2, using an FDR cut-off of 0.05 and a log2 fold change cut-off of 1.
Step 3: Extract the intersection genes that are both DEG and TF genes for further analysis, and then generate 01.TF.genes.exp.tsv (expression in TPM) based on this intersection. Genes with an average TPM of less than 0.5 are excluded.
@cyycyj
From the output of level assignment, you only got one gene that directly connects to the seed TF. It is because there is no TF genes connected to them for building a TO-GCN. So I will suggest you to use all TF genes (not only DEGs) to run the TO-GCN again.
If it still now works, you may add more TF genes to the seed set.
petitming
Dear Yao-ming,
Thank you for your advice. I tried the TO-GCN pipeline again, not only using the entire TF genes expression matrix (01.TF.genes.exp.txt), but also using all genes from MFSelector_Type1.txt (derived from the MFSelector pipeline) as the initial seed. However, the problem still remains: Node_relation.csv contains no output except the header, and Node_level.txt only contains the genes from MFSelector_Type1.txt.
I have attached the corresponding files, and I really appreciate your help.
01.TF.genes.exp.txt MFSelector_Type1.txt MFSelector_Type2.txt Node_relation.csv Node_level.txt
@cyycyj
I've tried the TO-GCN pipeline with your data and successfully got a TO-GCN with 8 levels. Here I used "C01G002267" as the seed only. The commands and output were listed below:
$./Cutoff 5 01.TF.genes.exp.txt No. of TFs: 1956 No. of time points: 5
Cutoff value for your reference: 0.94 ~ 0.98
$./TO-GCN 5 01.TF.genes.exp.txt seed.txt 0.9 No. of TFs: 1956 No. of time points: 5 No. of initial TF seeds: 1 Cutoff: 0.90
Assigning levels for nodes in GCN by Breadth-First-Search (BFS) method...... Done!
petitming
Dear Yao-ming,
Thank you for your debugging assistance. I discovered that my issue might be caused by the complex gene IDs. When I use the simplified version of the gene IDs, as I provided to you, it works well.
By the way, could you please provide me, and anyone else trying to use this epic tool, with a more detailed guide on how to build a network in Cytoscape based on Node_level.tsv
and Node_relation.csv
? I found several misleading methods for building Cytoscape networks on the internet.
Dear Yaoming,
First of all, I would like to express my sincere appreciation for your excellent tool, TO-GCN. It has been really useful for my research.
Today, while I was working with TO-GCN_STAR, something strange happened:
Do you have any idea about it? And please check the 01.TF.genes.exp.tsv file here, I hope it can be helpful for you.
Best regards,
Andrew
01.TF.genes.exp.txt