Support common featureCounts use cases: SAF format, long-read mode, per-read details

[jayhesselberth]

Summary

Based on research of common featureCounts use cases on GitHub, fcount-rs already supports the primary RNA-seq workflow well. This issue tracks three missing features to expand coverage:

Feature	Priority	Use Case
SAF format	High	ATAC-seq/ChIP-seq peak counting
Long-read mode (-L)	Medium	Nanopore/PacBio transcriptomics
Per-read details (-R)	Low	Debugging/QC

---

1. SAF Format Support

Why: SAF (Simplified Annotation Format) is essential for ATAC-seq and ChIP-seq workflows where users count reads in peaks rather than genes. This is the #3 most common use case on GitHub (after RNA-seq and paired-end counting).

SAF Format:

GeneID  Chr  Start  End  Strand
peak1   chr1  100    200  +
peak2   chr1  500    600  -

• Tab-separated, 5 columns
• Optional header row
• 1-based coordinates

Implementation:

• Auto-detect SAF format from file extension (.saf) or content (5 columns, no feature type)
• Add src/annotation/saf.rs parser
• Each SAF row becomes one Feature (peaks are standalone, no exon grouping)

---

2. Long-Read Mode (-L)

Why: Growing use of Nanopore/PacBio for transcriptomics (e.g., scNanoGPS for single-cell nanopore). The -L flag in featureCounts adjusts counting for long-read characteristics.

What -L does:

• Counts reads spanning multiple exons as single feature
• Adjusts overlap handling for reads that may span entire genes
• Handles the many CIGAR operations typical of long reads

Implementation:

• Add -L/--long-reads flag to CLI
• Adjust overlap resolution for long-read characteristics
• Consider minimum overlap as fraction of feature, not read

---

3. Per-Read Details Output (-R)

Why: Useful for debugging and understanding assignment decisions.

Implementation:

• Enable the existing -R/--details placeholder in CLI
• Output format: read_name status gene_id overlap_length
• Add src/output/details.rs writer

---

References

Common featureCounts usage patterns from GitHub:

• nf-core/atacseq - Uses featureCounts with SAF for peak counting
• CebolaLab/ATAC-seq - BED to SAF conversion for FRiP scores
• gaolabtools/scNanoGPS - featureCounts for nanopore transcriptomics