Report card for Correlation, Tree, and Seurat Mapping on Macaque motor cortex (Zemke et al. 2023)Permalink

OverviewPermalink

A taxonomy was initially built using the BICCN’s macaque motor cortex single nucleus 10x Multiome dataset. In building the taxonomy, 1000 binary marker genes were selected based on their gene expression from the single-cell transcriptome. Subsequently, the dataset was mapped to itself, termed self-projection, for evaluating the ideal performances of correlation, tree, and seurat mapping algorithms.

Quantitative analysisPermalink

The analysis evaluates the predictions of correlation, tree, and Seurat mappings in determining cluster labels in a self-projection evaluation.

Correlation MappingPermalink

1. Label-wise F1-score
   

2. Confidence values for correctly and incorrectly assigned labels
   

3. Confusion matrix (row-normalized)
   

Tree MappingPermalink

1. Label-wise F1-score
   

2. Confidence values for correctly and incorrectly assigned labels
   

3. Confusion matrix (row-normalized)
   

Seurat mappingPermalink

1. Label-wise F1-score
   

2. Confidence values for correctly and incorrectly assigned labels
   

3. Confusion matrix (row-normalized)