Cell Line¶
Use cell line commands to turn any common spelling of a cell line into a Cellosaurus accession, then read its identity, its cross-references, and its curated sequence variations.
Key boundaries:
- Cellosaurus is the only source. The CVCL accession is the join key for every other cell line dataset.
- Search compares normalized names: lowercase, with every non-alphanumeric character dropped.
MOLM13,MOLM-13, andMolm 13are the same query. - BioMCP ranks candidates and picks no winner. When two lines match exactly, both rows appear.
- The base card never fetches cross-references. Ask for
xrefswhen you need the join keys. - Every output names the Cellosaurus release and carries the CC BY 4.0 attribution line.
Search cell lines¶
QUERY is required. It can be a name, a synonym, or a CVCL accession.
biomcp search cell-line MOLM13
biomcp search cell-line "MV4;11"
biomcp search cell-line KG1 --limit 5
biomcp search cell-line CVCL_2119
Each row carries match (exact or partial) and, on exact rows, matched_on (name or synonym). Exact identifier matches sort first, then exact synonym-only matches, then partial matches. Human lines come first within each group.
A hyphenated query runs twice, once as typed and once with the hyphens removed, and the two windows merge by accession. KG-1 and KG1 therefore return the same set.
When Cellosaurus fills its 1000-row window, the total is unknown and the output carries a note. Use a more specific name or the accession.
Get cell line records¶
The card carries the accession, the RRID, the identifier name, synonyms, species, disease, category, sex, and age.
A source ID also works. BioMCP resolves it to the accession first and prints the ID it resolved from.
biomcp get cell-line ACH-000362 # DepMap
biomcp get cell-line SIDM00437 # Cell Model Passports
biomcp get cell-line CHEMBL3706573 # ChEMBL
biomcp get cell-line MOLM13_950_2019 # PharmacoDB
Request cell line sections¶
Cross-references, the join keys other datasets use:
Every key is printed. A line with no link to a resource shows an empty list, so biomcp get cell-line CVCL_0007 xrefs prints gdsc and cosmic_clp as empty.
Curated sequence variations:
Rows appear as Cellosaurus published them, with the HGVS description, the HGNC link, zygosity, and the PubMed sources. BioMCP adds no interpretation.
ChEMBL assay coverage:
The section prints the ChEMBL cell line record that names the accession: the ChEMBL ID, the EFO and CLO IDs, and how many ChEMBL assays were run on the line. The count tells you whether ChEMBL literature assays exist. BioMCP lists no assays and no activity values. The join goes through the accession alone, so biomcp get cell-line CVCL_0004 chembl finds the line ChEMBL spells K562.
PharmacoDB drug response:
The section counts the experiments PharmacoDB published for the line in each dataset. MOLM-13 has 1117 across CTRPv2, gCSI, GDSC1, and GDSC2. The join goes through the PharmacoDB ID in xrefs first and the line name second, and BioMCP keeps the answer only when the accession PharmacoDB returns is the accession asked for. The section lists no rows, because the unscoped listing for a heavily screened line such as K-562 is 19.4 MB of body.
The Cellosaurus sections together:
all covers variants and xrefs. The chembl and drug_response sections each cost their own request, so they are asked for by name.
Helper commands¶
One helper reads the PharmacoDB rows behind the drug_response counts, one dataset at a time:
--dataset is required. Without it the command fails before any request and names the ten PharmacoDB datasets, because an unscoped listing is megabytes of body. Each row names the compound, the experiment ID, and the published AAC, IC50, EC50, Einf, HS, and DSS1. PharmacoDB gives no units, BioMCP interprets no value, and an absent metric prints as -. A compound tested twice in one dataset stays two rows with two experiment IDs.
To go from a compound back to the lines it was tested on, run biomcp drug cell-lines <name>.
To go the other way, from a gene to the cell lines that express it, run biomcp gene cell-lines <symbol> --group <group>. That helper prints the Human Protein Atlas nTPM of one gene across one cancer group and carries the Cellosaurus accession on each row it could resolve by name.
JSON mode¶
JSON carries data_as_of and data_as_of_kind at the top level. data_as_of_kind is release when BioMCP read the Cellosaurus release, and retrieved when that read failed and the value is a retrieval time instead. A filled provider window appears in _meta.notes, and pagination.total is null.
Practical tips¶
Two names that look alike are often two lines. HL-60 is CVCL_0002; the NCI-60 line HL-60(TB) is CVCL_A794, a separate record. The NCI-DCTD link on HL-60(TB) carries the name HL-60, so a name-based join from NCI-60 lands on the wrong line. Join on the accession, not the name.
NB4 is the identifier of CVCL_0005 and also a synonym of the neuroblastoma line SJNB-4 (CVCL_8821). Both rows are exact matches. The matched_on column tells them apart.
KG1 matches three records, two of them mouse lines. Read the species column before you pick.