What is the difference between a heatmap-dendrogram of a gene-gene correlation matrix made from one cluster versus multiple clusters?
0
0
Entering edit mode
24 months ago
Pratik ▴ 840

Asking for anyone's insight on this please:

I think I need to be hit with this from multiple angles/perspectives for this to settle in my mind.

Background: I am analyzing a publicly available dataset of human fetal pancreas development weeks 12 through 22. I am currently looking at one time point (week 12). I used Seurat to cluster the 644 cells that make up the week 12 time point into 25 clusters(potentially different cell types). I am now interested in seeing the mechanistic insights that can be gleaned from one cell cluster (cell type) from the 25 clusters (multiple cell types).

What I did: I subsetted the data out of the cluster of interest (let's say: 35 cells that make up cluster 3). I then log 10000 normalized the UMI counts from cluster 3, and then extracted the top 1000 variable features/genes. Following this, I transposed the results of the top 1000 variable genes by 35 cells and then ran pearson correlation matrix on this, to result in a 1000 gene by 1000 gene correlation matrix. I then generated a heatmap-dendrogram of the matrix.

Furthermore, I did the same heatmap-dendrogram generation on the entire 644 cells that make up the week 12 time point (rather than just 35 cells of cluster 3 within the 644 cells of the week 12 time point).

I'm planning to loop through the dataset and do this heatmap-dendrogram generation for each cluster, and am hoping to glean cluster/cell-type-specific gene coregulation for GRN analysis. What do you think about this?

My question: What is the difference between a heatmap-dendrogram of a gene-gene correlation matrix generated from a cell type (cluster of cells (ie. 35 cells of cluster 3 within the 644 cells of the week 12 time point)) versus multiple cell types together all at once(a bunch of clusters together simultaneously (644 cells of week 12))?

A more specific question is if someone could please answer: What information is each providing?

I have included low resolution images of both heatmap-dendrograms:

Heatmap-dendrogram of top 1000 variable features by top 1000 variable features from cluster 3 enter image description here

Heatmap-dendrogram of top 1000 variable features by top 1000 variable features from all clusters in week 12 time point enter image description here

I may not be clearly understanding something here such as what is happening when I convert the genes by cells table to gene-gene pearson correlation matrix?

Very Respectfully, Pratik

scRNA-Seq Seurat RNA-Seq R rna-seq • 1.1k views
ADD COMMENT

Login before adding your answer.

Traffic: 1471 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6