Drug target evidence in single-cell data

Meta-analysis of drug target evidence in single-cell data

Contents

• analysis - Notebooks and scripts for analysis
• data - Metadata and output files (see Data Pointers)
• src - Main workflow scripts and functions
  • sc_target_evidence_utils - Python package with utility functions
• tests - Unit tests for utility functions

Set-up

# Make conda env (see also sc-target-evidence-env.yml)
conda create --name sc-target-evidence-env
conda activate sc-target-evidence-env

# Install R dependencies (for DE analysis)
conda install conda-forge::r-base==4.0.5 
ENVPATH=$(conda info --envs | grep sc-target-evidence-env | cut -d " " -f 5) # get path to conda environment
Rscript --vanilla -e "install.packages(c('BiocManager'), repos='http://cran.us.r-project.org', lib='${ENVPATH}/lib/R/library'); library('BiocManager'); BiocManager::install('glmGamPoi', lib='${ENVPATH}/lib/R/library')"
Rscript --vanilla -e "install.packages(c('tidyverse'), repos='http://cran.us.r-project.org', lib='${ENVPATH}/lib/R/library')"
Rscript --vanilla -e "library('BiocManager'); BiocManager::install('scater', lib='${ENVPATH}/lib/R/library')"

# install utils package
pip install .

Data pointers

Additional processed data is available via Figshare (doi:10.6084/m9.figshare.25360129)

Metadata

• suppl_table_diseases.csv - table of diseases available in CZ CellxGene database considered for study (supplementary table 1)
• cellxgene_hsapiens_donor_metadata.disease_relevant_annotation.csv - curated donor-level metadata for all adult studies in CxG collection, including grouping of diseases, grouping of tissues, annotation of disease-relevant tissues (supplementary table 2).
• target_universe_dict.json - dictionary of all the tested gene target universes used throughout analysis (built in analysis/make_target_universe.py)

scRNA-seq data

• [see figshare] cxg_aggregated_scRNA.tar.gz - AnnData objects of aggregated scRNA-seq data used for DE analysis for each disease. Gene expression counts are aggregated by sample and cell type annotation.

Diagnostic plots

Plot folders: sc_target_evidence/data/plots/{disease_id}_{disease_name}

• **cellxgene_{disease_id}.celltype_harmonization.*** - confusion table of original cell ontology annotations and uniformed ontology annotations. Heatmap color and number in cells denotes the number of cells for each category.
• **cellxgenetargets{disease_id}.n_cells_boxplot.*** - boxplot of numbers of cells per sample and cell type in healthy and disease tissue
• **cellxgenetargets{disease_id}.target_expression.*** - heatmap of log-normalized expression of a sample of drug targets for the disease
• **cellxgenetargets{disease_id}.celltype_distribution.*** - confusion table of assignment of uniformed cell type ontology to each donor, to check differences in cell type distribution across donors/datasets/diseases

Analysis outputs

• [see figshare] DEA_results.tar.gz - Results of differential expression analysis for each disease
• suppl_table_disease_target_evidence.csv - merged table of target-disease pairs with clinical status from OpenTargets, genetic evidence from OpenTargets and single-cell evidence from DE analysis, for all tested diseases.
• suppl_table_drugs.csv - merged table of drugs considered for analysis (investigational or approved drugs for analysed diseases).
• suppl_table_odds_ratios.all.csv - Results of association analysis between omic support (evidence) and clinical success (clinical status) across diseases
• suppl_table_odds_ratios.disease.csv - Results of association analysis between omic support (evidence) and clinical success (clinical status) by disease