analogous cell line types in human and mouse

[yongqunh]

This issue is relate to the CL issue #832: https://github.com/obophenotype/cell-ontology/issues/832 initially raised by @amadeusX, and discussed by @cmungall and @addiehl

The goal here is to compare analogous human and mouse cell line types, such as the pairs K562 and MEL, GM12878 and CH12.LX, and h1-hESC and CJ7 (this pair somewhat debated). It is important to evolutionarily compare human and murine cells. @amadeusX is interested in the evolutionary remodeling of transcription factor binding sites between diverse human and mouse primary cell or cell line cell pairs. See more details in the CL issue #832.

This poses an interesting topic for CLO. Some of the listed terms are in CLO, some not yet and need to be added. Then we need to see how to model them ontologically.

[yongqunh]

Three pairs: K562 cell: http://purl.obolibrary.org/obo/CLO_0007050 MEL cell: http://purl.obolibrary.org/obo/CLO_0037379 (newly added)

GM12878 cell: http://purl.obolibrary.org/obo/CLO_0022851
CH12.LX: https://web.expasy.org/cellosaurus/CVCL_0211 (murine lymphoma; need to add to CLO)

h1-hESC (=Line H1 cell): http://purl.obolibrary.org/obo/CLO_0037281 CJ7: https://web.expasy.org/cellosaurus/CVCL_C316 (synonym: ES-CJ7; CJ7-mESC; CJ-7; need to add to CLO if not yet)

[amadeusX]

Dear Jongqunh, Thank you so much! Is there any algorithmic way to find such pairs? It seems to be difficult for cell lines, but for primary cells, is there an algorithmic - programmable way to find comparable pairs? Thank you!

[yongqunh]

Hi Istvan, I would like to understand your question first. What kind of "comparable pairs" do you mean? Since the examples you provided are all cell line cells. Each cell line cell is derived from an in vivo cell type which is defined in Cell Ontology (CL). Therefore, a pair of two cell line cells are derived from a pair of two in vivo cell types, one from human, the other from mouse. Such information can be identified in CLO where we define the relation using 'derived from'. Once you find the pair of in vivo cell types, you will need to find whether they are from the same ancestor cell. In this case, what type of "ancestor" cell do you want to trace back to? Like epithelial cell? Also, one difference between cell line cell and in vivo cell is that the cell line cells are likely having new gene mutations so that they become immortal which cannot be found in the in vivo cells. Are you interested in the additional gene mutations? Oliver

[amadeusX]

Hi Jongqunh or Oliver, As you know, mouse is being used as a model for human, and mouse cells are being used as models for human cell types. Because of these models are very imperfect, scientists used the mouse models with much less validation than necessary, and the problems were swept under the carpet. I mean under "comparable pairs of cell types" where the mouse cell can be used as a tolerable model of the human cell. Practically, the mouse cell closest to the given human cell type, and importantly, the human cell closest to the mouse cell type. Originally, I called them analogous pairs but were told that there is no solid scientific background for analogy. This is a justified criticism.
The problem with immortalized cell lines is that they frequently diverge considerably from the originating cell types. Cancer cells may become more-or-less spherical, cell cycle and gene expression patterns run out of control, etc. This is, in part, is due to mutations, but I do not want to go that far. Thank you so much! Istvan

[yongqunh]

Thanks for your clarification. CLO does have a relation called 'is model for' as described in the original CLO paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4387853/ which may be applied. If you have a list of such comparable pairs, we may do some modeling and ontology representation, and see how it works out. It might be a good study.