Human Myeloma Cell Line (HMCL) NCATS MIPE 4.0 Drug Screen Dataset

Abstract Multiple myeloma, a hematopoietic malignancy of terminally differentiated B cells, is the second most common hematological malignancy after leukemia. While patients have benefited from numerous advances in treatment in recent years resulting in significant increases to average survival time following diagnosis, myeloma remains incurable and relapse is common. To help identify novel therapeutic agents with efficacy against the disease and to search for biomarkers associated with differential response to treatment, a large-scale pharmacological screen was performed with 1,912 small molecule compounds tested at 11 doses for 47 human myeloma cell lines (HMCL). Raw and processed versions of the drug screen dataset are provided, as well as supportive information including drug and cell line metadata and high-level characterization of the most salient features of each. The dataset is publicly available at Zenodo and the workflow code used for data processing and generation of supporting figures and tables can be found at https://github.com/khughitt/hmcl-drug-screen-pipeline. Overview The dataset shared here contains the raw and processed versions of a high throughput small moleculescreening dataset, as as supporting tables and figures. The computational pipeline used to generate the dataset is available at: https://github.com/khughitt/hmcl-drug-screen-pipeline Data organization The .zip file available on Zenodo contains the complete output from the computational pipeline linked to above. Data and figures are organized into subfolders relating to the various pipeline steps. Files & folders: cell_viability/ Average cell line viability table clusters/ Cell line and drug clustering results combined_viability_matrices/ All viability measurements combined into one row per drug/cell; used infer drug-drug and cell-cell similarity matrices drug_curves/ Table with one row per drug/cell line pair containing the curve fitting model input fit parameters drug_plates/ Drug screen viability measurements at different levels of processing, arranged with one column per plate drug_response_matrices/ Cell x drug matrices, with separate matrices reporting the viability measurements associated with each drug concentration, as well as matrices containing inferred AC-50 and log AC-50 values. fig/ Summary and quality assurance figures generated from the drug screen data metadata/ Cell line, drug, and plate-level metadata mutations/ Cell line mutation predictions with identifiers harmonized to match those used in the drug screen (source: Keats Lab) projections/ Cell and drug similarity matrix PCA and UMAP projections raw/ Raw version of the drug screen dataset, including the data for cell lines which were filtered out in the pipeline due to quality purposes similarity/ Cell-cell and drug-drug similarity matrices datapackage.yml Metadata for the provided data resources including sha256 checksums, human-readible descriptions, column types, etc. (see: https://datapackage.org/)