Add Cell Lines

[BAevermann]

We are in the process of curating Tahoe 100m and a number of public datasets which will require support for cell lines.

In addition, Idan, Jenny, and I have consolidated around a schema proposal.

---

Proposal (@gabdank)

1. Field Rename: tissue_type → sample_type

Change the primary field name from tissue_type to sample_type to better reflect the broader range of biological sample sources being curated
Idan created GitHub issue #1365

2. Enhanced Sample Type Categories (formerly known as tissue_type):

Addition of cell line as a new sample_type category
Rename cell culture → primary cell culture to distinguish from immortalized cell lines
Retain existing categories: tissue, organoid

3. Add in_vitro_cell_ontology_term_id:

Required for primary cell culture: Must use CL (Cell Ontology) terms
Required for cell line: Must use Cellosaurus identifiers
NA for tissue and organoid types

4. Update the logic for tissue_ontology_term_id:

Required when sample_type is tissue or organoid and is UBERON,
NA for cell line and primary cell culture types

---

An alternative proposal that would save us from adding new property.

1. Field Rename: tissue_type → sample_type

Change the primary field name from tissue_type to sample_type to better reflect the broader range of biological sample sources being curated
Idan created GitHub issue #1365

2. Field rename tissue_ontology_term_id -> sample_ontology_term_id

Required when sample_type is tissue or organoid and Must use UBERON terms,
Required for primary cell culture: Must use CL (Cell Ontology) terms
Required for cell line: Must use Cellosaurus identifiers

3. Enhanced Sample Type Categories (formerly known as tissue_type):

Addition of cell line as a new sample_type category
Rename cell culture → primary cell culture to distinguish from immortalized cell lines
Retain existing categories: tissue, organoid

[jahilton]

For consideration, the known cell line Datasets in CELLxGENE are these

• https://cellxgene.cziscience.com/collections/d0e9c47b-4ce7-4f84-b182-eddcfa0b2658
  • ca421096-6240-4cee-8c12-d20899b3e005 - H1299
  • 4b9e0a15-c006-45d9-860f-b8a43ccf7d9d - Calu-3
• https://cellxgene.cziscience.com/collections/00109df5-7810-4542-8db5-2288c46e0424
  • fe2479fd-daff-41a8-97dc-a50457ab1871 - MCF7
  • 38a78e91-e361-45c8-bae3-1d6d5108d18e - K562
  • 108c2b65-4ea9-4792-8b5b-451842faf35b - A549

[brianraymor]

There would be some additional effort for cellosaurus per our current standards for the cellxgene ontology guide:

1. OWL is unavailable. @gabdank has indicated that the creators suggest XML+XSD rather than the OBO.

Q11: Is there a difference in the content of Cellosaurus in the structured text (TXT), OBO and XML formats?
Yes, you should be aware that:
The OBO format file (cellosaurus.obo) does not contain: the STR profile data, the age at sampling and the full reference records (authors, title, journal). All the comments are concatenated.
And while the TXT and XML versions of Cellosaurus have the same data content, there are some differences in terms of the organization of the data. These differences are described below.
The TXT version is split into two data files: cellosaurus.txt and cellosaurus_refs.txt, the first file contains all Cellosaurus information except for the full reference records which are stored in the second file. This is not the case for the XML version which consists of a single data file, cellosaurus.xml and a XML schema definition file, cellosaurus.xsd
In the TXT file cross-references to external resources do not contain the URLs of these cross-references, these can be instantiated using the information stored in the file cellosaurus_xrefs.txt

2. There are no versioned/tagged releases? The FTP download references the latest release.
3. Term deprecation is documented in a separate file - cellosaurus_deleted_ACs.txt List of deleted accession numbers/entries rather than using annotation for obsolete terms. Are there cases where terms are replaced by new terms?

CL example - also see Principle: Stability of Term Meaning (principle 19)

[Term]
id: CL:0000002
name: obsolete immortal cell line cell
def: "OBSOLETE: A cell line cell that is expected to be capable of an unlimited number of divisions, and is thus able to support indefinite growth/propagation in vitro as part of a immortal cell line." [ReO:mhb]
comment: Obsoleted in July 2013 and replaced by the CLO 'immortal cell line cell' class, as a result of CLO-OBI-CL alignment efforts.\n\nCovers cells actively being cultured or stored in a quiescent state for future use.
synonym: "continuous cell line cell" EXACT []
synonym: "permanent cell line cell" EXACT []
is_obsolete: true
replaced_by: CLO:0000019

4. Term identifiers do not use OBO-format identifiers, meaning a CURIE (prefixed identifier) of the form Ontology:Identifier. Compare with the CL example above.

[Term]
id: CVCL_KA96

[brianraymor]

RE the Cell LIne Ontology (CLO) - see Is CLO still maintained?

[gabdank]

Regarding cllosaurus:

The release version of the XML file is in the header:

<header>
  <terminology-name>Cellosaurus</terminology-name>
  <description>Cellosaurus: a knowledge resource on cell lines</description>
  <release version="52.0" updated="2025-04-10" nb-cell-lines="163868" nb-publications="29267"/>
</header>

[brianraymor]

Thanks for the reference - but it's impossible for the ontology table to reference a stable url for a specific download for the cellxgene ontology guide to process. All other ontologies have tagged releases in github.

Yes - the table can specify the "2025-04-10" release as documentation. And COG also keeps a copy of the ontology based on an earlier experience with the *Dv ontologies evaporating. But I'm noting the exception ...

[jychien]

@brianraymor Here is a spreadsheet of the comparison of CLO, EFO, BTO, and Cellosaurus based off of the Cellosaurus API: ontology_comparison. To note:

• I did spot check and found the information to be accurate.
• These are the cell lines found in 100M Tahoe plus a few other common cell lines
• The ones marked in red denote redundancy or cell lines that are not found in the ontology

[brianraymor]

Hi @jychien - is there further guidance related to final ontology selection?

CLO may be eliminated because it is not meeting the maintenance and responsiveness principles for obofoundry per my earlier comment.

EFO almost meets the baseline requirements (1 missing term)?

Are there additional advantages to Cellosaurus that would outweigh its costs for special processing ?

[gabdank]

Cellosaurus represents modified cell lines. For example, K562 cell line captured by a single EFO term http://www.ebi.ac.uk/efo/EFO_0002067, but in Cellasaurus, there are: 361 terms representing the various modifications of K562 cell line (https://www.cellosaurus.org/search?query=K562), in addition to the basic cell line (https://www.cellosaurus.org/CVCL_0004). The ability to capture a specific modified cell line ID is valuable for perturbation experiments utilizing cell lines.

[brianraymor]

@gabdank - for cellosaurus, are the names/labels included in datasets/presentation or only the term identifiers? I noted that some appear a bit random or esoteric:

[Term]
id: CVCL_X345
name:#822

...

[Term]
id: CVCL_C4ZN
name:#G65G

Is there nomenclature guidance?

[gabdank]

I think that the names will appear in the datasets, and the ones you are pointing out are indeed esoteric, as these are hybridoma cell lines constructed for antibody production, so unlikely to be part of single-cell analysis.