search

wehos / CellT

An offical repo for paper "Single-cells are Spatial Tokens: Transformers for Spatially Resolved Transcriptomics Data Imputation"
14 stars 2 forks source link

Dataset Issue #1

Open yikang613 opened 1 year ago

yikang613 commented 1 year ago

Hi, thank you for sharing your valuable work. I noticed that the preprocessing method in the utils.py file expects the data to be in the .h5ad format. However, the dataset I downloaded from the website referenced in your paper is in the .csv format. I'm wondering if there's a step where the .csv file needs to be converted into a .h5ad file, or perhaps I might have downloaded the incorrect dataset. Could you please clarify this?

wehos commented 12 months ago

Hi! Thank you for your interest. You are correct that the data need preprocessing to be transformed to an h5ad file. We intended to release the processed data, as we mentioned, preprocessed data will be provided in this repo after official release of the paper. Unfortunately, the paper is now under revision and we will release a new version in November hopefully.

yikang613 commented 11 months ago

I've processed the datasets and matched the rmse results from your paper. However, I couldn't reproduce the ari and nmi results due to the absence of "cell types" for the Lung datasets. The README didn't provide this information either. Could you guide me to where you sourced the cell type details from the website?

wehos commented 11 months ago

You should be able to find their official annotations from the Seurat object released by CosMx.