search

beyondpie / CEMBA_wmb_snATAC

Whole mouse brain snATAC seq analysis
MIT License
23 stars 4 forks source link

cellType cov weighted loss function is ambiguous #18

Closed GitHUB-ZYD closed 3 months ago

GitHUB-ZYD commented 3 months ago

image hello. i am ambiguous while read the paper with the loss function. And no code in directory 11.deeplearning.

What data was used to calculate the covariance matrix between different cell types? The y_true which used to calculate the loss ? the y_true of each epoch? Or y_true of the entire dataset ? shold diagonal elements need to be removed to calculate the mean of CoV for each cell type?

beyondpie commented 3 months ago

@GitHUB-ZYD Hi there, sorry for the late response.

  1. Please check the deep learning codes here: https://github.com/yal054/mba_dl_model. I will later upload the deep learning codes in this repo.

  2. For your questions, I will invite my colleague, who implemented the deep learning model, to answer them. I will also talk with him, and address the questions directly here. Thanks!

Songpeng

GitHUB-ZYD commented 3 months ago

I have found the answer in the code https://github.com/yal054/mba_dl_model. Thank you