search

tanjimin / C.Origami

C.Origami, a prediction and screening framework for cell type-specific 3D chromatin structure.
60 stars 9 forks source link

The pre-trained models of other cell types #32

Open lhy0322 opened 9 months ago

lhy0322 commented 9 months ago

Dear Jimin, Your research has been very helpful in my work! I am trying to reproduce the best results of your method as our baseline to compare. Does Origami provide pre trained models for other cell types? (excluding IMR90)

thanks！

tanjimin commented 9 months ago

Hi @lhy0322 , sorry for the late reply. The idea of C.Origami is that the model trained on IMR-90 will also work on different cell types as well as shown in the paper. You can preprocess the ATAC-seq and CTCF ChIP-seq from other cell types and try the model directly.

May-0707 commented 1 week ago

Dear Jimin, I preprocessed the ATAC-seq and CTCF CUT&Tag from mm10 and try the model. The default epoch paramater is 80. The model saved at epoch=79-step=41760.ckpt does not produce any graphical results, whereas the model at epoch=0-step=522.ckpt can generate some graphs, though it is still far inferior to your pre-trained model. The email I sent you described the picture in detail. HiC-pro and hicpro2higlass.sh generated mcool, cool2npy.py change it to npz. CUT&Tag preprocessed with https://yezhengstat.github.io/CUTTag_tutorial/. Can you give me some suggestions for improvement or just use your pre-traind model ? Best regards, May Young