search

yutongo / TIGON

MIT License
31 stars 3 forks source link

How can I derive the results of GRN? #2

Open wyqhyrz opened 4 months ago

wyqhyrz commented 4 months ago

Hi, Yutong Thanks for building this fantastic tool. I am trying to apply this tool to my projects. However, I am not familiar with algorithms. I have some questions about the application of this tool. How can I derive the results of GRN? I need genes that play a key role in the process of change. Using "Training.ipynb", I just got the file "ckpt.pth", however, I don't know how to process this file to get more information I need.

Best, Yuqian.

shenscore commented 2 months ago

same question

cadyyuheng commented 1 month ago

I encountered the same issue. After reviewing their paper, it seems the Jacobian of velocity is what corresponds to the 'Generate the plot of regulatory matrix' step in their visualization notebook. In the EMT example, they trained on 3D input (UMAP1, 2, 3), which resulted in a 3x3 Jacobian matrix. I assume that if we trained on k-dimensional gene data, we would obtain a kxk matrix, which would be more interpretable as a gene regulatory network (GRN). This is my current understanding.