Package to annote binding type of bioactivity measures based on keyword search of
- abstracts from PubMed, PubChem assay description, CrossRef or Google Patents
- assay descriptions from ChEMBL assay descriptions
The annotation is currently supported for two types of targets
- Class A GPCRs with a 3-level hierarchical keyword search to annotate compounds as orthosteric, allosteric, bitopic, covalent or unknown based work in Burggraaf et al., J. Chem. Inf. Model. (2020)
- Protein Kinases with 1-level keyword search to annotate compounds as allosteric or unknown based (extended) keywords from Christmann-Franck et al., J. Chem. Inf. Model. (2016)
pip install git+https://github.com/sohviluukkonen/BindingType.git@main
The package has both an API and a CLI which can process either
- Papyrus datasets
- lists of document and/or assay IDs
In the case of Papyrus-dataframe, the annotation will a new BindingType column to the dataframe and can be done from the command line with
bindtype_papyrus -i <dataset.csv/.tsv> -tt <GPCR/Kinase>
or with the API with
from bindtype.papyrus import add_binding_type_to_papyrus
df = add_binding_type_to_papyrus(df, target_type=GPRC/Kinase)
There is also an option to annotate all 'unknown' compounds that based on their Tanimoto similarity to the annotated compounds: -sim, --similarity flag in the CLI and similarity=True in the API.
In the more general case, the annotation will create dictionaries based list of document IDs and/or assays IDs. This can be done either from the command line with
bindtype -did <document_id_file_path> -aid <assay_id_file_path> -tt <GPCR/Kinase>
or with the API with
# for the GPCRs
from bindtype import ClassA_GPCR_HierachicalBindingTypeAnnotation
parser = ClassA_GPCR_HierachicalBindingTypeAnnotation()
# for the kinases
from bindtype import Kinase_AllostericAnnotation
parser = Kinase_AllostericAnnotation()
# Only abstracts
dct_doc_annotations = parser(document_ids=list_of_document_ids)
# Only assay descriptions
dct_assay_annotations = parser(assay_ids=list_of_assay_ids)
# Both
dct_doc_annotations, dct_assay_annotations = parser(document_ids=list_of_document_ids, assay_ids=list_of_assay_ids)
As the scripts were developed with data from Papyrus and uses document and assay description IDs should be in the format used in the all_doc_ids and AID columns: PMID:<pubchem_id>, PubChemAID:<pubchem_assay_id>, DOI:, PATENT:<patent_id> and <chembl_assay_id>.