gefeiwang
commented
1 year ago Hi Cristal,
We used slice 6 (from 0 to 8) for pie charts visualization. It is ok to use the entire scRNA-seq dataset since we do not have a scRNA-seq dataset for each slice, and the entire dataset should contain most major cell types appear in the ST data.
Best, Gefei
cristalliao
Dear Professor,
I have a question about comparing it with other deconvolution methods. In the paper, there is one figure about the Estimated proportions of cell types obtained from STitch3D (j) and other compared methods (k) in a 6.5 PCW heart slice visualized by a pie chart in Figure 5j and Figure 5k.
I want to know about which slices you used for this figure in the human heart dataset (slice0-8).
Also, when I use the slice0 human heart dataset in CARD. The input needs single cell RNAseq (scRNA-seq) count data, along with meta information indicating the cell type information and the sample (subject) information for each cell. Can I use the whole single-cell RNAseq for all slices like the figures below:
Because as you mentioned, there is not a single-cell RNA-seq reference dataset for slice0 specifically. We can use the entire scRNA-seq dataset as the cell-type reference data for all slices.
Does it make sense? Then I can obtain the cell type deconvolution results using the CARD method and compare them with the STitch3D:
The CARD method
The STitch3D method
Could you help me to look at if it is a correct analysis? Thanks in advance!