HiCAT utilizes subset markers structured in a 3-level taxonomy tree, major-types, minor-types and subsets.
Although the three levels of cell-type taxonomies roughly correspond to the lineage tree, we rather define them by their transcriptomic profiles for classification and clustering purposes.
- Major-types are defined as cell populations that are clearly separable from others so that they are easily identifiable in a low-dimensional landscape, e.g., in tSNE or UMAP plots,
- Minor-types are a subpopulation of a major-type and not clearly separable even though they can be well localized within a major-type cluster.
- Subsets are a subpopulation of a minor-type and typically represents specific activation/polarization states of a minor population, e.g., M1/M2 macrophage, Th1,2,17 of CD4+ T cells. They are not clearly separable from other subsets and also may not be well localized such that blind clustering is not applicable to catch a specific subset.
- Rather than create a completely new markers db, try to use the existing one, to which you can add new types with appropriate markers (in Hugo symbols).
- When you add a new major-type, other than those in the existing major-types, it must be clearly separable (e.g. in t-SNE plot) from others.
- If your new types are not clearly separable from existing major types, it must be added as a minor-type or a subset of an existing major-type that it belongs or is most close to. You may check what major-type is suitable to your minor-type using t-SNE or UMAP plot.
- Typically, minor-types from the same major-type (or subsets from the same minor-type) share some marker genes. You can add the common markers to all the minor-types (or subsets) that share.
- It is recommended that you first run HiCAT using the existing markers db. Later, you can add new types with an appropriate markers as minor-types of, for example, epithelial cells or fibroblast.