Skip to content

🧬 Reference Genomes

MosaiCatcher supports multiple reference genome assemblies. Each assembly requires specific normalization files, blacklist regions, and a matching BSgenome R package for StrandPhaseR.

Supported Assemblies

Assembly Organism Status Container tag
hg38 (GRCh38) Human Fully supported hg38-v2.4.0
hg19 (GRCh37) Human Fully supported hg19-v2.4.0
T2T (CHM13v2) Human Fully supported T2T-v2.4.0
mm10 (GRCm38) Mouse Fully supported mm10-v2.4.0
mm39 (GRCm39) Mouse Fully supported (see caveats) mm39-v2.4.0
canfam3 Dog Framework-ready (normalization files pending) —
canfam4 Dog Framework-ready (normalization files pending) —

Selecting a Reference

Set the reference parameter in your config or via --config:

snakemake --cores <N> --config data_location=<PATH> reference=mm39 --sdm conda

For containerized runs, the container image is automatically selected to match the reference value:

snakemake --cores <N> \
    --config data_location=<PATH> reference=hg19 \
    --sdm conda apptainer \
    --apptainer-args "-B /g:/g"
# Uses ghcr.io/friendsofstrandseq/mosaicatcher-pipeline:hg19-v2.4.0

Assembly-Specific Notes

mm39 (GRCm39) Caveats

mm39 is fully supported but has two known limitations to be aware of:

Missing 50 kb / 100 kb normalization files

Normalization files for 50 kb and 100 kb bin sizes are not yet available for mm39. Use window: 200000 (200 kb bins) when analyzing mm39 data:

# config/config.yaml
reference: mm39
window: 200000

HGSVC blacklist — chrX validation needed for XX samples

The blacklist regions file for mm39 was generated using an XO (male) control cell line. As a result, chrX coverage normalization may be biased for XX (female) samples. We recommend:

  • Excluding chrX from analysis for XX samples: chromosomes_to_exclude: [chrX]
  • Or interpreting chrX SV calls from XX mm39 samples with extra caution

mm39 chrX results

For XX (female) mouse samples on mm39: chrX blacklist regions were derived from a male (XO) control. SV calls on chrX should be validated independently.

canfam3 / canfam4 (Dog)

The reference genome infrastructure (genome download, chromosome lists) is in place for canfam3 and canfam4. However, normalization files and blacklist regions are not yet generated. These assemblies are framework-ready and will be fully supported in a future release.

Pre-built BWA Indexes

To skip building BWA indexes locally, enable the download_prebuilt_indexes parameter:

download_prebuilt_indexes: True

This downloads pre-built indexes from AWS iGenomes. Useful on HPC systems where index building is time-consuming or storage-constrained.

Shared Reference Directory (HPC)

On HPC systems with multiple users, configure reference_base_dir to point to a shared location:

reference_base_dir: /g/korbel/shared/genomes

All users sharing this directory avoid redundant downloads and index building.