What problem does it solve? Counting transposable elements alongside genes from RNA-seq BAMs is error-prone: multimapping reads need special handling, strandedness must be verified per library, and featureCounts version drift breaks reproducibility. This Skill turns nf-core/rnaseq star_salmon BAMs plus a pre-built grouped TE SAF into integer gene + TE subfamily count matrices inside a version-locked Docker container (te-fc:2.0.2, featureCounts v2.0.2). ## Core Features & Use Cases - Env-locked two-pass counting: A vendored driver runs a TE pass (SAF, -M integer Random-One, no --fraction) and a gene pass (GTF, per-library strandedness), then row-binds them into a combined matrix. - Docker staging wrapper: scripts/run_te_counting.sh symlinks BAMs into bam_fin/ and bind-mounts targets at identical host=container paths so symlinks resolve and column names come out clean. - Strand-split QC suite: qc/run_qc.sh runs a 3-pass -R CORE regime witness, closure audit, -O silent-loss attribution, young-family gate, and geometry concordance, printing a consolidated GREEN/RED verdict. - Use Case: You have finished an nf-core/rnaseq run on mouse mm39 samples and need a combined gene + TE subfamily count matrix ready for DESeq2/edgeR differential expression, with verified integer counts and a QC gate before handoff. ## Quick Start Run scripts/run_te_counting.sh with your BAM directory, gene GTF, grouped TE SAF, verified gene strandedness, and output directory to produce the gene, TE, and combined count matrices inside the locked te-fc:2.0.2 container.
