scCirclehunter

Step1 - Identification of ecDNA from scATAC-seq based on a pseudo-bulk algorithm (circlehunter2)

Installation

micromamba create -f scCirclehunter.yaml
micromamba activate scCirclehunter
python setup.py install

Usage

Circlehunter2 can also be used for bulk ATAC-seq if the -b option is not specified. If the -b CB option is provided, read counts for each ecDNA segment will be output and saved in an .h5ad file. Since similar operations are performed in the downstream analysis of Step 2, we recommend running the following command directly for both bulk and single-cell ATAC-seq. The result of the following command can be found in GBM4349_circlehunter2_ecDNA.bed.

circlehunter2 -p 16 --blacklist /share/references/hg38/blacklist/hg38.blacklist.sorted.bed /mnt/2w/data2/andy/scATAC-Seq/rawdata/cellranger/GBM4349/outs/possorted_bam.bam /home/andy/Projects/circlehunter2/workspace/dev/data/GBM4349_circlehunter2_ecDNA.bed

input BAM file have to be coordinate sorted, index of the BAM file should present with a .bai suffix, more options please check

circlehunter2 -h

Output

cirlcehunter2 output an extended BED format file with columns:

1. chrom: contig name of which the segment come from
2. start: segment start
3. end: segment end
4. name: No. of this segment, format as ecDNA_{ecDNA_no}_{total_segments}_{segment_no}, you can use these info to form a complete ecDNA circle
5. score: no data in this columns, always a .
6. strand: orientation of this segment
7. start_ci: confidence interval of segment start
8. end_ci: confidence interval of segment end
9. start_peak: discordant enrich region of segment start
10. end_peak: discordant enrich region of segment end
11. start_cross: reads counts of which cross the segment start breakpoint and disjoint the breakpoint
12. end_cross: reads counts of which cross the segment end breakpoint and disjoint the breakpoint
13. linked_reads: reads counts of which connected this segment and the next one
14. depth_mean: mean of depth of this region, maybe false positive if this is smaller than cutoff[^cutoff]
15. high_coverage: fraction of region that depth is higher than cutoff, maybe false positive if this is pretty small

[cutoff]: cutoff is a user input params -c or determine by circlehunter2 automatically if not set by user, which is default to the inverse survival of poisson distribution use mean of the whole genome depth as $\lambda$. cutoff will be log in the stderr.

Step2 - Assigning ecDNA to Cell Populations

Refer to the example in demo/scCirclehunter_downstream_demo.html for assigning ecDNA to cell populations. The functions utilized are available in the scripts directory.