[Design pattern] Adding a new cell type for every anatomical structure in Uberon

[scheuerm]

There are some members of the CL community that are proposing to add new cell types (e.g., endothelial cell subtypes) for every defined anatomical structure (every different blood vessel in Uberon). Without having any evidence for a distinct characteristic, I do not think that the location in a given anatomical structure should be sufficient for defining a new cell type in CL.

[scheuerm]

There is a debate about whether cells present in a unique anatomic location should be considered cells of a unique type. I would argue that being present in a unique location, in and of itself, does not necessarily mean that it is a unique type of the thing and that a type should be an intrinsic property of the thing, regardless of its location and its environment.

Let’s use a different example to make the argument. The CD3E protein is a part of the T cell receptor complex and involved in signal transduction in T cells. Is the CD3E protein in a helper T cell a different type of protein from a CD3E protein in a cytotoxic T cell? Is a central memory T cell CD3E a different type of proteins from an effector memory T cell CD3E? For the sake of argument, let’s imagine that the pH of the cytoplasm of the helper T cell is lower than the pH of the cytoplasm of the cytotoxic T cell, and as a result, the cytoplasmic tail of CD3E in helper T cells adopts a different conformation (conformation B) 90% of the time. I would argue that confirmation A and confirmation B are two different types of CD3E. And while confirmation B is found in helper T cells it is not the location that dictates the different CD3E type, but rather the intrinsic property (conformation B) that is different. Indeed, there are still 10% of the CD3E proteins in the helper T cells that are conformation A. Taken to the extreme, is actin in a T cell a different type of actin from actin in a B cell? Should we expect the Protein Ontology to represent a different type of protein for every different type of cell it is located in?

Thus, just because a cell is in a different location does not, in and of itself, mean that it is a different type of cell. If a property of the cell is different as a result of the different location and if this is a shared property of cells in that location, then yes, it should be considered a different type of cell. However, without evidence to indicate that the cell’s properties are distinct, it should not be given a unique designation as a type in the Cell Ontology, just because in theory it could be different. And while subject matter experts may be well equipped to judge whether this is the case or not, their opinion, without empirical evidence, is not sufficient.