kidcancerlab/rrrSingleCellUtils

Single cell analysis utility functions for the Ryan Roberts lab

Roberts Lab R Single Cell Analysis Utility functions

These are utility functions for use by the Roberts lab to analyze single cell sequencing data.

---

Functions

Working and tested functions

• tenx_load_qc()
  • Load 10X data
  • Formerly tenXLoadQC()
    • Now requires mt_pattern and species_pattern instead of spec
• gen_cellecta_bc_data()
  • Extract cellecta barcode information from a sam or bam file
• plot_complex_heatmap()
  • Make a complex heatmap showing nichenet output from FindLigands()
• find_ligands()
  • Formerly findLigands()
  • Run nichenetr to find ligands potentially inducing receptor-driven gene expression changes
• find_tar_genes()
  • Formerly findTarGenes()
  • Create a gene list containing putative targets of ligand activity
• kill_cc()
  • Formerly killCC()
  • Regress out cell cycle effects
• plot_cc()
  • Plot the proportion of cells in each cell cycle stage
• process_raw_data
  • This is a wrapper for several functions to get and process single cell data from the NCH IGM core. I have built the defaults to be specific to the Roberts lab, so you may need to carefully change the defaults if you want to use it outside of this context. The input is a link to data and a sample sheet that outlines the information about each sample (see inst/exampleSampleInfoSheet.txt - Column headers must remain unchanged). The data are then downloaded using smbclient and then md5sum checked and untar’d. The data are then processed with cellranger mkfastq and either cellranger count or cellranger-dna cnv (depending on the exp_type argument).

Still in testing

• process_ltbc()

To do

• Make arguments inherit documentation from other functions instead of re-typing them
• Add argument to provide path to cellranger
• Add argument to specify bcl2fastq path or module
• Vignette
• Function to check sample sheet
• For multiomics, make sure both atac and gex are present in sample input sheet, and not as "sample1_atac" and "sample1_gex", which will fail. Should be "sample1" and "sample1" for both.
• Try out snp calling: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1863-4
• In tenx_load_qc, check if all data is filtered out or sobj is empty
• Figure out multiomic FRiP calculation and see if I need to divide by 2
  • https://support.10xgenomics.com/single-cell-multiome-atac-gex/software/pipelines/latest/algorithms/overview
• Change make_sobj to be a sbatch script
• in getProcessData.R, change sbatch submissions to use use_sbatch_template()
• Make optimize_silhouette warn if there are no dim reductions or clustering present
• Try out snp calling: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1863-4
• In tenx_load_qc, check if all data is filtered out or sobj is empty
• Change make_sobj to be a sbatch script
• in getProcessData.R, change sbatch submissions to use use_sbatch_template()

---

Errors

If you run into any errors, please run the following commands and send the output to me along with the error messages produced.

traceback()
sessionInfo()