Revised 0.16.0 build (2 Jul 2026): two additive revisions were folded into the binaries below. The self-contained HTML results report was redesigned (branded single file, zoomable SVG figures, separate up- and down-regulated gene tables, functional-enrichment tables, and per-step runtimes), the figures step now also produces separate up- and down-regulated top-DEG heatmaps, and Check Environment probes every optional pipeline route (Trim Galore, Trimmomatic, SortMeRNA, RiboDetector, FastQ Screen, RSeQC, UCSC utilities) and the R analysis packages. The default STAR → featureCounts → DESeq2 analysis path is byte-identical to the original 0.16.0 (the Fusarium graminearum baseline reproduces exactly at 2,723 up / 2,478 down).
BulkSeq Studio 0.16.0 — a large additive feature release. The default STAR → featureCounts → DESeq2 analysis path is byte-identical to 0.15.x (the Fusarium graminearum baseline reproduces exactly at 2,723 up / 2,478 down), so existing results are unchanged; every new capability below is opt-in.
Added
- Alternative differential-expression engines. Alongside DESeq2 (default), RNA-seq counts can be tested with limma-voom or edgeR (quasi-likelihood F-test). All three emit the same result schema, so enrichment, networks, and figures are identical regardless of engine. Concordance with DESeq2 is high (log2FC Spearman 0.997–1.000, top-DEG Jaccard 0.94–0.97, 100% direction agreement; benchmark B14).
- Alternative read trimmers. Selectable fastp (default), Trim Galore, or Trimmomatic, each with its own parameters in the GUI.
- Alternative rRNA removal. SortMeRNA (default, reference-based) or RiboDetector (reference-free, CPU, no database).
- Contamination screening. Optional FastQ Screen step maps a read subsample against a panel of reference genomes.
- Single-end FASTQ support. Single-end libraries now run the full alignment route (trimming, rRNA filtering, STAR/HISAT2/Salmon, featureCounts). Mixed single/paired layouts are rejected with guidance.
- GSVA pathway activity. Optional per-sample gene-set activity scoring from your own custom gene sets (organism-safe), with a heatmap.
- RSeQC alignment QC. Optional read-distribution and gene-body-coverage reports (genome-BAM routes).
- Self-contained HTML report. A single-file
results_report.htmlinlining figures, top genes, enrichment, and provenance; reachable from the GUI. - Guided design/covariate builder. A dialog composes the DESeq2 design formula from your metadata columns (batch/covariate adjustment) without typing R.
- Advanced parameters panel. A collapsible per-tool section exposing the important parameters of each tool, with the validated defaults pre-filled.
Changed
- Pipeline environments gained edger, trim-galore, cutadapt, pigz, trimmomatic, fastq-screen, bowtie2, ribodetector, gsva, rseqc, and the UCSC gtfToGenePred/genePredToBed utilities; the pinned lock was regenerated (CPU ONNX runtime, no CUDA).
- Validation extended to a human dataset (airway smooth muscle ± dexamethasone, GRCh38/Salmon): the pipeline recovers the canonical glucocorticoid signature (FKBP5, ZBTB16, KLF15, SPARCL1 up; VCAM1 down) — benchmark B15.
Removed
- htseq-count dropped as a quantifier option (featureCounts, STAR gene counts, and Salmon/tximport cover the same ground).
Downloads
- Windows installer:
BulkSeqStudio-Setup-0.16.0.exe— per-user install, no admin rights. - Windows portable:
BulkSeqStudio-Portable-0.16.0.zip— unzip and run. - Linux AppImage:
BulkSeqStudio-0.16.0-x86_64.AppImage—chmod +xand run (built on Ubuntu 24.04; needs glibc 2.39+). - Linux portable:
BulkSeqStudio-linux-0.16.0.tar.gz— extract and run the bundled launcher.
The validation benchmark archive (families B1–B15) is on Zenodo: https://doi.org/10.5281/zenodo.20955660
Per-asset SHA-256 checksums are shown next to each file in the Assets section below.