Right now the muon ATAC module is designed around fragment files from 10x-style scATAC-seq workflows. But a growing number of users (including myself) work with bulk ATAC-seq count matrices — peaks × samples — that they want to bring into the muon/MuData ecosystem for multi-omics integration.
There's no clean supported path for this. You can manually construct an AnnData object but there's no ac.read_bulk_atac() or equivalent that handles the expected metadata (peak annotations, FRiP/TSS QC metrics as .obs columns, consensus peak BED coordinates as .var).
Would be useful to have a thin ingestion utility similar to how sc.read_csv() works — just with ATAC-specific .var fields pre-populated from a BED file and .obs QC columns mapped from a MultiQC summary.
I can draft an implementation if there's appetite for this.
Right now the muon ATAC module is designed around fragment files from 10x-style scATAC-seq workflows. But a growing number of users (including myself) work with bulk ATAC-seq count matrices — peaks × samples — that they want to bring into the muon/MuData ecosystem for multi-omics integration.
There's no clean supported path for this. You can manually construct an AnnData object but there's no ac.read_bulk_atac() or equivalent that handles the expected metadata (peak annotations, FRiP/TSS QC metrics as .obs columns, consensus peak BED coordinates as .var).
Would be useful to have a thin ingestion utility similar to how sc.read_csv() works — just with ATAC-specific .var fields pre-populated from a BED file and .obs QC columns mapped from a MultiQC summary.
I can draft an implementation if there's appetite for this.