Broad_Babel
Minimal name translator of JUMP consortium.
Installation
pip install broad-babel
Broad sample to standard
You can fetch a single value. Note that only ORF datasets have an associated broad_id by default.
from broad_babel.query import broad_to_standard
broad_to_standard("ccsbBroad304_99994")
If you provide multiple strings it will return dictionary.
broad_to_standard(("ccsbBroad304_09930", "ccsbBroad304_16164"))
Wildcard search
You can also use sqlite bindings. For instance, to get all the samples that start as "poscon" you can use:
from broad_babel.query import run_query
run_query(query="poscon%", input_column="pert_type", output_columns="JCP2022,standard_key,plate_type,pert_type", operator="LIKE")
Make mappers for quick renaming
This is very useful when you need to map from a long list of perturbation names. The following example shows how to map all the perturbations in the compound plate from JCP id to perturbation type.
from broad_babel.query import get_mapper
mapper = get_mapper(query="compound", input_column="plate_type", output_columns="JCP2022,pert_type")
Export database as csv
from broad_babel.query import export_csv
export_csv("./output.csv")
Custom querying
The available fields are:
- standard_key: Gene Entrez id for gene-related perturbations, and InChIKey for compound perturbations
- JCP2022: Identifier from the JUMP dataset
- plate_type: Dataset of origin for a given entry
- NCBI_Gene_ID: NCBI identifier, only applicable to ORF and CRISPR
- broad_sample: Internal Broad ID
- pert_type: Type of perturbation, options are trt (treatment), control, negcon (Negative Control), poscon_cp (Positive Control, Compound Probe), poscon_diverse, poscon_orf, and poscon (Positive Control).
You can fetch any field using another (note that the output is a list of tuples)
run_query(query="JCP2022_915119", input_column="JCP2022", output_columns="broad_sample")
It is also possible to use fuzzy querying by changing the operator argument and adding "%" to out key. For example, to get the genes in the "orf" dataset whose name start with "RBP"(some of which are retinol-binding proteins) we can do:
[x[:2] for x in run_query(
"RBP%",
input_column="standard_key",
output_columns="standard_key,JCP2022,plate_type",
operator="LIKE",
) if x[2]=="orf"]
Note that we also got RBPMS here, which is actually RNA-binding protein with multiple splicing, so use this with caution.
Additional documentation
Metadata sources and additional documentation is available here.
Note that Babel only contains metadata of JUMP compounds and genes, and may not contain sample information from other projects (e.g., LINCS). A more comprehensive table to map "broad ids" to standard chemical ids (e.g., SMILES, InChiKey) can be found here.