An overview of the CellRanger-ARC Pipeline Analysis

Strategies for cell recovery when regular Cell RangerARC pipeline fails to count cells.
In this talk we cover:
Cell RangerARC, Cell RangerARC-reanalyze, Cell Ranger-count and Cell Ranger-count atac

Transcript

1. Cellranger-ARC pipeline and cell recovery strategies Cynthia Soto Cardinault Oct 11, 2024
2. A suite of analysis pipelines for processing Chromium multiome scATAC and scRNA sequencing data which enables detailed analysis of both data-types, and their interconnections. Cell Ranger ARC Workflow: One sample, one GEM well, one ATAC + one GEX flow cell
3. Cell Ranger ARC pipeline: components ATAC Matrix Computer: ATAC FASTQs → Peak-Barcode Matrix GEX Matrix Computer: GEX FASTQs → Gene-Barcode Matrix Associates a subset of barcodes observed to enables data analysis at single cell resolution to identify the cell population from the non-cell background. FILTERING STEPS: • Barcodes must pass 3 filtering criteria. The first 2 are `ATAC exclusions`. • If a barcode fails any of the filtering criteria is masked from the total set of barcodes prior to cell calling. ALGORITHM STEPS: • Deduplication step • Ordmag-derived initial grouping • K-means boundary refinement. • Map classification to de-duplicated barcodes.
4. Cell Ranger ARC Command run
5. Cellranger ARC: Outputs Cellranger ARC reanalyze Outputs summary metrics for GEX and ATAC
6. Cellranger ARC: Outputs Cellranger ARC reanalyze Outputs summary metrics for GEX and ATAC
7. Join Cell Calling Possible causes affecting the counts? What resources do we have to recover cells? 1. Cell caller override: we can select the minimum of transposition event and UMIs per cell to include. 1. Barcode selection: we need to identify our own barcodes, to run a secondary analysis with the cellranger-arc reanalyze command
8. Cell caller override Cell caller override: we can select the minimum of transposition event and UMIs per cell to include
9. Two additional parameters need to be provided to cellranger-arc count: --min-atac-count=N (minimum number of transposition events in peaks for a cell barcode) --min-gex-count=N (minimum number of UMI counts for a cell barcode) Cell caller override
10. --barcodes=LIS T Specify barcodes to use in analysis. The barcodes could be specified in a text file that contains one barcode per line (blank lines are ignored). CSV (with/without a header) is also accepted. Only the first column of the CSV is used — exports from Loupe Browser will have this format. Required if neither --peaks nor --params has been specified. Barcode selection 1. Barcode selection: we need to identify our own barcodes, to run a secondary analysis with the cellranger-arc reanalyze command The command reruns secondary analysis performed on the peak-barcode matrix (dimensionality reduction, clustering and visualization) using different parameter settings. cellranger-arc reanalyze --id=my-sample_reanalysis \ --barcodes=valid_barcodes.csv \ --matrix=/my-sample/outs/raw_feature_bc_matrix.h5 \ --reference=/refs/hg19 \ --atac-fragments=/my-sample/outs/atac_fragments.tsv.gz
11. Barcode selection You need to have valid `cellranger-arc` outputs You need to use `cellranger-arc reanalyze` command You need to provide a list of valid `--barcode=list.csv` Pipeline output only acts over the secondary analysis. It means in the output/analysis directory
12. Identification of valid barcodes GEX Barcode translation Matching cells `Cellranger count` GEX ATAC Multiome raw data `Cellranger-atac count` ATAC Barcode translation GEX barcodes GEX barcodes
13. Cell Ranger ARC Reanalyze Output
14. Recovery cells example cellranger-arc reanalyze --id=my-sample_reanalysis \ --barcodes=valid_barcodes.csv \ --matrix=/my-sample/outs/raw_feature_bc_matrix.h5 \ --reference=/refs/hg19 \ --atac-fragments=/my-sample/outs/atac_fragments.tsv.gz
15. Considerations and thinks about Review library quality control metrics. Recovered cells do not equate to high quality. Cellranger-arc is more stringent than Cellranger's standalone pipelines. Cellranger pipelines apply different algorithms, so results vary between pipelines. Some quality control metrics depend on tissue type, particularly frozen tissue is challenging. Review Cell Ranger ARC Troubleshooting.
16. Thanks