[SonderXiaoming]

SMARTer smRNA libraries need protocol-specific trimming and must bypass the Total RNA UMI workflow. This change adds a selectable smRNA profile while preserving smarter_total_rna as the default, and produces a separate concise RNA-seq summary alongside the comprehensive report.

Changes

• Add --kit smarter_smrna for single-end R1 or paired-end R1/R2 input, with sense/forward orientation by default, structure/poly(A) trimming, Q20 filtering, and no UMI extraction or deduplication.
• Make STAR profiles configurable through Bash arrays; route downstream QC, expression, circRNA inputs, coverage, and reporting through the selected profile. Retain chimeric output and disable biological poly(A)-tail detection for smRNA libraries.
• Include kit/effective strand in step metadata and reject incompatible cross-kit resume attempts. Document that v1.0 output signatures require regeneration.
• Generate a standalone Step 27 summary of read retention, mapping, genomic regions, orientation, and top genes; support gene-only expression matrices when circRNA results are absent.
• Update README, test/reproducibility documentation, synthetic fixtures, release manifest, and checksums. Normalize text to LF for reproducible hashes across platforms.

Validation

• bash tests/smoke/run_smoke.sh: passed on Ubuntu with Python 3.12, including syntax/CLI checks, synthetic smRNA preprocessing and alignment routing, STAR argument injection, cross-kit resume rejection, and 4 summary-report tests.
• sha256sum -c repro/checksums.sha256: all 57 files passed; manifest hashes/sizes and staged Git blobs were checked as well.
• git diff --cached --check: passed.
• The smRNA smoke branch uses test doubles for cutadapt, STAR, and samtools. Optional plotting/EMapper checks were skipped locally because their Linux dependencies were unavailable. The full reference-based workflow, production alignment/circRNA accuracy, and comprehensive R Markdown rendering have not been validated in this change.

Data handling

Only source, documentation, and synthetic sequencing fixtures are included. Public configuration retains placeholder paths; local configuration, sample output directories, and test work files are ignored. Staged contents were checked for credentials, private paths, private identifiers, and generated outputs. The new commit uses the contributor's GitHub noreply email.

Draft pending contributor review.