bioflowkit 0.2.0__tar.gz → 0.2.1__tar.gz

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/.github/workflows/candidate-smoke-test.yml

@@ -17,6 +17,11 @@ permissions:
   contents: read
   pull-requests: write   # for the summary comment
 
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch so
+# the v4/v5 actions stop emitting Node 20 deprecation warnings.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
 jobs:
   smoke:
     name: Validate + smoke-test changed candidates

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/.github/workflows/ci.yml

@@ -6,6 +6,11 @@ on:
   pull_request:
     branches: [main]
 
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch so
+# the v4/v5 actions stop emitting Node 20 deprecation warnings.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
 jobs:
   unit-tests:
     name: Unit tests (Python ${{ matrix.python-version }})
@@ -47,12 +52,11 @@ jobs:
         run: ruff check .
 
   typecheck:
-    # Advisory only — mypy issues do not block merging.  Provides
-    # visibility while we gradually add type annotations.  Will be
-    # upgraded to a blocking check when the error count reaches 0.
-    name: Type check (mypy, advisory)
+    # Blocking: the bioflow package type-checks clean under mypy
+    # (--ignore-missing-imports for the un-stubbed docker / anthropic /
+    # openai SDKs).  Keep it that way — a new type error fails CI.
+    name: Type check (mypy)
     runs-on: ubuntu-latest
-    continue-on-error: true
 
     steps:
       - uses: actions/checkout@v4
@@ -66,7 +70,7 @@ jobs:
         run: pip install -e ".[dev]" && pip install types-PyYAML types-requests
 
       - name: Run mypy
-        run: mypy bioflow --ignore-missing-imports || true
+        run: mypy bioflow --ignore-missing-imports
 
   registry-schema:
     name: Validate registry YAMLs
@@ -108,12 +112,12 @@ jobs:
           EOF
 
   digest-audit:
-    # Advisory until enough tools are pinned — prints the missing-digest
-    # count and lists the first ~30 unpinned tools.  Upgrade to a blocking
-    # check (drop `|| true`) once the bulk of the registry is pinned.
-    name: Container digest pin audit (advisory)
+    # Now blocking: every *active* tool must carry an image_digest
+    # (deprecated tools are skipped — their upstream images are gone by
+    # definition).  This keeps the registry fully content-addressed so a
+    # silently-retagged or GC'd upstream image can never change results.
+    name: Container digest pin audit
     runs-on: ubuntu-latest
-    continue-on-error: true
 
     steps:
       - uses: actions/checkout@v4
@@ -126,5 +130,5 @@ jobs:
       - name: Install dependencies
         run: pip install pyyaml
 
-      - name: Count missing image_digest entries
-        run: python scripts/pin_digests.py --audit || true
+      - name: Require every active tool to be digest-pinned
+        run: python scripts/pin_digests.py --audit

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/.github/workflows/docs.yml

@@ -26,6 +26,11 @@ concurrency:
   group: pages
   cancel-in-progress: true
 
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch so
+# the v3/v4/v5 actions stop emitting Node 20 deprecation warnings.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
 jobs:
   build:
     runs-on: ubuntu-latest

bioflowkit-0.2.1/.github/workflows/nfcore-concordance.yml

@@ -0,0 +1,75 @@
+name: nf-core concordance
+
+# Manually-dispatched concordance benchmark: run bioflow and the matching
+# nf-core pipeline on a golden dataset, then score their agreement with
+# scripts/compare_nfcore.py.
+#
+# This is NOT a per-PR gate.  A real run needs tens of GB of references
+# and hours of compute, so it expects a runner that already has them
+# staged (a self-hosted runner, or a large GitHub runner the maintainer
+# provisions).  It is run deliberately before a release and its JSON
+# output is published with the release notes.
+#
+# See docs/benchmarks/nfcore-concordance.md for the datasets, method, and
+# acceptance thresholds.
+
+on:
+  workflow_dispatch:
+    inputs:
+      comparison:
+        description: "Which comparison to run"
+        type: choice
+        options: [vcf, counts]
+        default: vcf
+      bioflow_output:
+        description: "Path to the bioflow output (VCF or counts TSV) on the runner"
+        required: true
+      reference_output:
+        description: "Path to the nf-core output on the runner"
+        required: true
+      threshold:
+        description: "Min Jaccard (vcf) or Spearman rho (counts) to pass"
+        required: false
+        default: "0.90"
+
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
+jobs:
+  score:
+    name: Score concordance
+    # The default GitHub runner cannot hold iGenomes references; point this
+    # at a self-hosted runner that has them staged.  Left as ubuntu-latest
+    # so the workflow is valid and the *scoring* step is exercisable on
+    # pre-staged outputs.
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Score bioflow vs nf-core
+        run: |
+          if [ "${{ inputs.comparison }}" = "vcf" ]; then
+            python scripts/compare_nfcore.py vcf \
+              --bioflow "${{ inputs.bioflow_output }}" \
+              --reference "${{ inputs.reference_output }}" \
+              --out concordance.json \
+              --min-jaccard "${{ inputs.threshold }}"
+          else
+            python scripts/compare_nfcore.py counts \
+              --bioflow "${{ inputs.bioflow_output }}" \
+              --reference "${{ inputs.reference_output }}" \
+              --out concordance.json \
+              --min-rho "${{ inputs.threshold }}"
+          fi
+
+      - name: Upload concordance report
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: nfcore-concordance
+          path: concordance.json

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/.github/workflows/nightly-smoke.yml

@@ -10,6 +10,11 @@ on:
     - cron: "0 3 * * *"
   workflow_dispatch:
 
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch so
+# the v4/v5 actions stop emitting Node 20 deprecation warnings.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
 jobs:
   smoke-matrix:
     name: Recipe smoke matrix

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/.github/workflows/release.yml

@@ -22,6 +22,11 @@ on:
 permissions:
   contents: read
 
+# Opt into the Node 24 runtime ahead of the 2026-06-16 forced switch so
+# the v4/v5 actions stop emitting Node 20 deprecation warnings.
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+
 jobs:
   build:
     name: Build sdist + wheel

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/CHANGELOG.md

@@ -12,6 +12,189 @@ ship bug fixes only.  Breaking changes to the documented public API
 
 ---
 
+## [0.2.1] — 2026-06-12
+
+> **Why upgrade from 0.2.0**: the `v0.2.0` tag predated the registry
+> freshness fix, so the 0.2.0 wheel bundled a registry whose
+> `quay.io/biocontainers/*` tags had been garbage-collected from Quay —
+> `rnaseq_deg`, `chip_seq`, `atac_seq`, `eukaryote_assembly`,
+> `metagenome_assembly`, and others would fail at run time with "image
+> not found".  **0.2.1 ships the repaired, 107/107-pinned registry**, so
+> recipes actually pull their containers.  Everything below landed after
+> the 0.2.0 cut and is new to PyPI users here.
+
+### Changed — nightly smoke matrix expanded (3 → 5 recipes)
+- Added real-container smoke cases for **chip_seq** (TrimGalore — a
+  container family shared by chip/atac/methylation, previously
+  unexercised) and **germline_variants** (validates the variant recipe's
+  fastp wiring).  All five pass against real BioContainers locally
+  (~90 s) — more of the "522 mostly-mock tests" critique closed with
+  genuine end-to-end coverage.
+
+### Changed — mypy type-checking is now blocking
+- Fixed all 33 mypy errors across 7 modules (Rich `TaskID` optionals in
+  the progress bars, `dict[str, Any]` run kwargs, anthropic/openai/docker
+  SDK union + missing-stub noise, a loop-variable shadow in
+  `cli/update.py`).  `mypy bioflow --ignore-missing-imports` now reports
+  **0 errors**.
+- CI's `typecheck` job drops `continue-on-error` — a new type error now
+  fails the build, closing the "advisory only" gap.
+
+### Added — nf-core concordance benchmark (harness + methodology)
+- `scripts/compare_nfcore.py` (new, stdlib-only): scores agreement
+  between a bioflow output and the matching nf-core output —
+  **Jaccard + genotype concordance** on normalised VCF sites
+  (vs nf-core/sarek), and **Spearman ρ** of per-gene counts
+  (vs nf-core/rnaseq).  Optional `--min-jaccard` / `--min-rho` gate for
+  CI.
+- `docs/benchmarks/nfcore-concordance.md`: golden datasets
+  (GIAB HG002 chr20; nf-core/rnaseq chr22), method, and initial
+  acceptance thresholds.
+- `.github/workflows/nfcore-concordance.yml`: manually-dispatched job
+  (not a per-PR gate — a full run needs staged references).
+- Honesty note: bioflow ships the *scoring* half (committed + tested);
+  the *production* half (running both pipelines on a machine with the
+  references) is operator-run and documented.
+- Tests: +13 (`tests/unit/test_compare_nfcore.py`).
+
+### Added — GPU passthrough + Podman runtime
+- `@stage(..., gpu=True)` (and a tool YAML's `resources.gpu`) now attach
+  all host GPUs to that stage's container via a Docker `DeviceRequest`
+  (the API equivalent of `--gpus all`); needs the NVIDIA Container
+  Toolkit, and degrades to a warning on CPU-only hosts rather than
+  failing.  Threaded through `Stage`, the `@stage` decorator, and the
+  preset `run_plan` path.
+- `DockerBackend` works with **Podman**: it honours `BIOFLOW_DOCKER_HOST`
+  / `DOCKER_HOST` (point it at the Podman API socket) and an optional
+  `base_url`, and reads `BIOFLOW_CONTAINER_RUNTIME`.
+- `bioflow doctor` recognises Podman as a Docker alternative — the
+  `docker_cli` / `docker_daemon` checks fall back to `podman` and report
+  which runtime they found.
+- Tests: +10 (`tests/unit/test_gpu_podman.py`).  Backend `run()` gains a
+  `gpu` kwarg (the `ContainerBackend` protocol + MockBackend updated).
+
+### Added — reference-DB catalog expansion + refgenie manifest
+- `bioflow/core/db.py` catalog gains the references real recipes need:
+  GATK known-sites `dbsnp_grch38` + `mills_indels_grch38` (BQSR/VQSR),
+  `encode_blacklist_grch38` (ChIP/ATAC peak filtering), and
+  `gencode_grch38` (STAR/Salmon/featureCounts annotation).
+- Catalog entries now carry `genome` + `asset` tags, and a new
+  `refgenie_manifest()` / `bioflow db manifest` emits a
+  [refgenie](https://refgenie.databio.org/)-compatible JSON mapping
+  `<genome>/<asset>` → catalogued DB, so labs already standardised on
+  refgenie can see which existing assets satisfy a bioflow requirement.
+- Tests: +5 (`tests/unit/test_db.py`).
+
+### Added — rnaseq_deg depth: GO enrichment + MultiQC
+- `rnaseq_deg` extended from 4 → 6 stages:
+  - **`enrich_go`** — GO enrichment on the significant DEGs via gseapy's
+    Enrichr query.  Symbol-based, so no organism-specific OrgDb package
+    is needed; an awk numeric-regex guard on the `padj` column avoids the
+    classic "NA-treated-as-0" false positive.
+  - **`multiqc_report`** — aggregates every per-sample fastp + Salmon
+    report into a single MultiQC HTML.
+- DAG-shape test updated; recipe still runs end-to-end under MockBackend.
+
+### Added — `joint_genotyping` recipe (GATK cohort best practice)
+- New 7-stage recipe (`bioflow/recipes/variant_calling/joint_genotyping.py`)
+  implementing the canonical GATK **joint-genotyping** workflow for
+  cohorts, where `germline_variants` only does single-sample direct
+  calling:
+  - **per sample (fan-out)**: fastp → BWA-MEM → MarkDuplicates →
+    HaplotypeCaller `-ERC GVCF`
+  - **cohort (converge)**: CombineGVCFs → GenotypeGVCFs →
+    best-practice hard filtering (separate SNP / INDEL filters) → SnpEff
+- Takes a `sample_id,fastq_r1,fastq_r2` sample sheet and uses `.starmap`
+  to run the per-sample stages in parallel before converging on the
+  joint steps — the production pattern reviewers expect for population
+  and family studies, and a worked example of bioflow's fan-out.
+- Recipe count 19 → 20 (12 per-pipeline).  Tests: +3 (registration, DAG
+  shape, MockBackend e2e).
+
+### Added — run provenance (RO-Crate + PROV-style JSON)
+- `bioflow/core/provenance.py` (new): every recipe run records, per
+  stage, the container **image + content digest**, the exact
+  **command**, every **input file's SHA-256 + size**, **start/end**
+  timestamps, exit code, and the bioflow version.
+- At the end of a run the workspace gains two self-describing files:
+  - `provenance.json` — flat, human-readable run record
+  - `ro-crate-metadata.json` — an [RO-Crate 1.1](https://www.researchobject.org/ro-crate/)
+    research object (the de-facto packaging standard for computational
+    workflow runs), so the output directory is consumable by reviewers
+    and downstream tools directly.
+- Wired into `bioflow recipe run` (on by default; `--no-provenance` to
+  skip).  Opt-in and **zero-cost when off** — the SDK hot path pays
+  nothing unless a recorder is installed, and provenance errors degrade
+  to warnings rather than aborting the science.
+- New `bioflow provenance show <workspace>` command (+ `--json`) renders
+  the recorded run: per-stage image, pinned digest, exit status, and
+  input hashes.  Builds directly on the digest-pinning work — pinned
+  tools show their `sha256:…` in the provenance.
+- Tests: +14 (`tests/unit/test_provenance.py`); verified end-to-end
+  against real Docker (digest resolved from the local image, RO-Crate
+  validates structurally).
+
+### Fixed — registry freshness (stale BioContainer tags)
+- **Discovery**: an audit during digest-pinning found that ~half the
+  registry's `quay.io/biocontainers/*` image tags had 404'd.  Quay
+  rotates each package's `--<buildhash>_<n>` build suffix and garbage-
+  collects the old ones, so dozens of recipes (chip_seq, atac_seq,
+  eukaryote_assembly, metagenome_assembly, rnaseq_deg's Salmon stage,
+  …) would have failed at run time with "image not found" — unrelated
+  to the user's data.
+- `scripts/refresh_tags.py` (new): audits every Quay BioContainer
+  reference, and with `--apply` rewrites any dead tag to the newest
+  *same-version* build (never changing the upstream software version).
+  Non-Quay images and versions that have left Quay entirely are
+  reported for manual review.
+- Applied: **34 registry tool YAMLs** + **7 recipe-hardcoded images**
+  (bowtie2, bracken, flye, macs3, medaka, metabat2, tobias) re-pointed
+  to live tags.  Salmon's `1.10.3--hb950928_0` → `1.10.3--h45fbf2d_5`
+  fixed separately (it broke the rnaseq_deg quant stage).
+- Verified: the bumped images pull + run (e.g. `bowtie2 2.5.4`), and
+  the nightly smoke matrix is green.
+
+### Added — full digest pinning (registry now 100% content-addressed)
+- Digest coverage raised from 5/110 → **107/107 active tools** in two
+  passes.  The second pass resolved the 17 hold-outs:
+  - **Version bumps** to the newest BioContainer of the same tool where
+    the pinned version had left Quay: DESeq2 1.44→1.50.2, edgeR
+    4.2→4.8.2, limma 3.60→3.66, clusterProfiler 4.12→4.18.4, topGO
+    2.56→2.62, HOMER 4.11.1→5.1, Scanpy 1.10.1→1.7.2 (1.10 was never on
+    Quay biocontainers), Comet 2024020→2026011, Percolator 3.06.1→3.7.1,
+    InterProScan 5.67→5.59.
+  - **Image-source switches**: methylKit and monocle3 moved off the
+    multi-GB `bioconductor/bioconductor_full:RELEASE_3_18` (gone) to the
+    dedicated `bioconductor-methylkit` / `r-monocle3` BioContainers;
+    Cell Ranger tag `7.2.0`→`v7.2.0`; Seurat `5.0.1`→`5.0.0`.
+  - **Deprecations**: MaxQuant joins MSFragger + FragPipe as
+    `deprecated: true` — proprietary tools whose images are gone and
+    which the `proteomics_dda` recipe does not use (it runs the
+    open-source msconvert → Comet → Percolator stack).  The audit now
+    skips deprecated tools.
+- **CI `digest-audit` is now blocking** (was advisory): every active
+  tool must carry an `image_digest`, so the registry can never silently
+  rot again.
+- `scripts/refresh_tags.py` also reports `version_gone` and non-Quay
+  images for these manual cases.
+
+### Added — Bioconda recipe (prep)
+- `conda-recipe/meta.yaml`: noarch-python Bioconda recipe (only
+  bioflow's pure-Python stack is a conda dep; tools run as Docker
+  containers).  Submission walkthrough in `docs/MAINTAINER.md` Part 7.
+  Gated on the real-PyPI publish (the recipe sources the PyPI sdist).
+
+### Changed — CI
+- All workflows opt into the Node 24 runtime
+  (`FORCE_JAVASCRIPT_ACTIONS_TO_NODE24`) ahead of the 2026-06-16 forced
+  switch, silencing the Node 20 deprecation warnings.
+
+### Fixed — nightly smoke
+- `rnaseq_deg.qc_one` smoke assertion matched the stage's real output
+  names (`<sample_id>_R1.clean.fq.gz`), fixing a false CI failure.
+
+---
+
 ## [0.2.0] — 2026-06-05
 
 First PyPI release.  Three months of 0.1.x work consolidated into a

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bioflowkit
-Version: 0.2.0
+Version: 0.2.1
 Summary: bioflow: one-line comparative-genomics recipes + Tier-A SDK, orchestrated over per-tool Docker BioContainers.
 Project-URL: Homepage, https://github.com/hope9901/bioflow
 Project-URL: Repository, https://github.com/hope9901/bioflow
@@ -35,7 +35,7 @@ Description-Content-Type: text/markdown
 [![PyPI](https://img.shields.io/pypi/v/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
 [![Downloads](https://img.shields.io/pypi/dm/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
 [![python](https://img.shields.io/pypi/pyversions/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
-[![tests](https://img.shields.io/badge/tests-565%20passed-brightgreen)](tests/)
+[![tests](https://img.shields.io/badge/tests-609%20passed-brightgreen)](tests/)
 [![nightly smoke](https://github.com/hope9901/bioflow/actions/workflows/nightly-smoke.yml/badge.svg)](https://github.com/hope9901/bioflow/actions/workflows/nightly-smoke.yml)
 [![license](https://img.shields.io/badge/license-MIT-lightgrey)](LICENSE)
 [![docs](https://img.shields.io/badge/docs-mkdocs-blue)](https://hope9901.github.io/bioflow/)
@@ -51,14 +51,14 @@ and a privacy-first LLM companion is available when you want it.
 
 ## What you get
 
-- **19 cookbook recipes** invokable as one-liners:
+- **20 cookbook recipes** invokable as one-liners:
   - *Comparative genomics (8)*: pangenome, ANI, phylogeny, GWAS,
     gene-family evolution, AMR/VF catalogue, COG enrichment, NCBI
     download.
-  - *Per-pipeline (11)*: prokaryote_assembly, eukaryote_assembly,
+  - *Per-pipeline (12)*: prokaryote_assembly, eukaryote_assembly,
     rnaseq_deg, metagenomics_profile, metagenome_assembly, scrna_seq,
     chip_seq, atac_seq, methylation_wgbs, proteomics_dda,
-    germline_variants.
+    germline_variants, joint_genotyping (GATK cohort best practice).
 - **110 tools** registered across 16 categories, all pulled as
   BioContainer images at run time — nothing to install on the host
   beyond Docker + Python.
@@ -66,6 +66,10 @@ and a privacy-first LLM companion is available when you want it.
   `runnable_slow` / `incompatible` against your CPU / RAM / GPU / arch.
 - **Input-hash caching**: re-running a recipe with unchanged inputs
   returns in seconds.
+- **Run provenance**: every recipe writes `provenance.json` +
+  `ro-crate-metadata.json` recording input SHA-256, container image
+  digests, commands, and timestamps — a self-describing research object
+  for reproducibility and journal submission.  `--no-provenance` to skip.
 - **Privacy-first LLM companion** (optional): terminology Q&A, sanitized
   error diagnosis, tool-registration assist.  Disabled by default.
 
@@ -202,19 +206,30 @@ Some pipelines need external databases (Pfam, eggNOG, BUSCO, etc.).
 
 ```bash
 bioflow db list                                  # show available DBs
-bioflow db fetch busco_bacteria --dest /refs
-bioflow db verify busco_bacteria --dest /refs
+bioflow db fetch dbsnp_grch38 --dest /refs
+bioflow db verify dbsnp_grch38 --dest /refs
+bioflow db manifest --dest /refs                 # refgenie-compatible asset map
 ```
 
-| Key | Size | Used by |
-|---|---:|---|
-| `busco_bacteria`   | 0.07 GB | busco |
-| `busco_insecta`    | 0.08 GB | busco |
-| `busco_vertebrata` | 0.30 GB | busco |
-| `pfam`             | 0.50 GB | interproscan |
-| `dfam_curated`     | 2.00 GB | repeatmasker, earlgrey |
-| `uniprot_sprot`    | 0.25 GB | braker3 |
-| `eggnog`           | 8.50 GB | eggnog_mapper |
+| Key | Size | Used by | refgenie asset |
+|---|---:|---|---|
+| `busco_bacteria`          | 0.07 GB | busco | — |
+| `busco_insecta`           | 0.08 GB | busco | — |
+| `busco_vertebrata`        | 0.30 GB | busco | — |
+| `pfam`                    | 0.50 GB | interproscan | — |
+| `dfam_curated`            | 2.00 GB | repeatmasker, earlgrey | — |
+| `uniprot_sprot`           | 0.25 GB | braker3 | — |
+| `eggnog`                  | 8.50 GB | eggnog_mapper | — |
+| `kraken2_standard_8gb`    | 7.50 GB | kraken2, bracken | — |
+| `bowtie2_grch38_noalt`    | 3.50 GB | bowtie2 | hg38/bowtie2_index |
+| `dbsnp_grch38`            | 1.60 GB | gatk4 | hg38/dbsnp |
+| `mills_indels_grch38`     | 0.02 GB | gatk4 | hg38/known_indels |
+| `encode_blacklist_grch38` | 0.001 GB | macs3, tobias | hg38/blacklist |
+| `gencode_grch38`          | 0.05 GB | star, salmon, subread | hg38/ensembl_gtf |
+
+`bioflow db manifest` emits a [refgenie](https://refgenie.databio.org/)-compatible
+JSON mapping `<genome>/<asset>` → catalogued DB, so a lab already on
+refgenie can see which existing assets satisfy a bioflow requirement.
 
 ---
 

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/README.md

@@ -3,7 +3,7 @@
 [![PyPI](https://img.shields.io/pypi/v/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
 [![Downloads](https://img.shields.io/pypi/dm/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
 [![python](https://img.shields.io/pypi/pyversions/bioflowkit.svg)](https://pypi.org/project/bioflowkit/)
-[![tests](https://img.shields.io/badge/tests-565%20passed-brightgreen)](tests/)
+[![tests](https://img.shields.io/badge/tests-609%20passed-brightgreen)](tests/)
 [![nightly smoke](https://github.com/hope9901/bioflow/actions/workflows/nightly-smoke.yml/badge.svg)](https://github.com/hope9901/bioflow/actions/workflows/nightly-smoke.yml)
 [![license](https://img.shields.io/badge/license-MIT-lightgrey)](LICENSE)
 [![docs](https://img.shields.io/badge/docs-mkdocs-blue)](https://hope9901.github.io/bioflow/)
@@ -19,14 +19,14 @@ and a privacy-first LLM companion is available when you want it.
 
 ## What you get
 
-- **19 cookbook recipes** invokable as one-liners:
+- **20 cookbook recipes** invokable as one-liners:
   - *Comparative genomics (8)*: pangenome, ANI, phylogeny, GWAS,
     gene-family evolution, AMR/VF catalogue, COG enrichment, NCBI
     download.
-  - *Per-pipeline (11)*: prokaryote_assembly, eukaryote_assembly,
+  - *Per-pipeline (12)*: prokaryote_assembly, eukaryote_assembly,
     rnaseq_deg, metagenomics_profile, metagenome_assembly, scrna_seq,
     chip_seq, atac_seq, methylation_wgbs, proteomics_dda,
-    germline_variants.
+    germline_variants, joint_genotyping (GATK cohort best practice).
 - **110 tools** registered across 16 categories, all pulled as
   BioContainer images at run time — nothing to install on the host
   beyond Docker + Python.
@@ -34,6 +34,10 @@ and a privacy-first LLM companion is available when you want it.
   `runnable_slow` / `incompatible` against your CPU / RAM / GPU / arch.
 - **Input-hash caching**: re-running a recipe with unchanged inputs
   returns in seconds.
+- **Run provenance**: every recipe writes `provenance.json` +
+  `ro-crate-metadata.json` recording input SHA-256, container image
+  digests, commands, and timestamps — a self-describing research object
+  for reproducibility and journal submission.  `--no-provenance` to skip.
 - **Privacy-first LLM companion** (optional): terminology Q&A, sanitized
   error diagnosis, tool-registration assist.  Disabled by default.
 
@@ -170,19 +174,30 @@ Some pipelines need external databases (Pfam, eggNOG, BUSCO, etc.).
 
 ```bash
 bioflow db list                                  # show available DBs
-bioflow db fetch busco_bacteria --dest /refs
-bioflow db verify busco_bacteria --dest /refs
+bioflow db fetch dbsnp_grch38 --dest /refs
+bioflow db verify dbsnp_grch38 --dest /refs
+bioflow db manifest --dest /refs                 # refgenie-compatible asset map
 ```
 
-| Key | Size | Used by |
-|---|---:|---|
-| `busco_bacteria`   | 0.07 GB | busco |
-| `busco_insecta`    | 0.08 GB | busco |
-| `busco_vertebrata` | 0.30 GB | busco |
-| `pfam`             | 0.50 GB | interproscan |
-| `dfam_curated`     | 2.00 GB | repeatmasker, earlgrey |
-| `uniprot_sprot`    | 0.25 GB | braker3 |
-| `eggnog`           | 8.50 GB | eggnog_mapper |
+| Key | Size | Used by | refgenie asset |
+|---|---:|---|---|
+| `busco_bacteria`          | 0.07 GB | busco | — |
+| `busco_insecta`           | 0.08 GB | busco | — |
+| `busco_vertebrata`        | 0.30 GB | busco | — |
+| `pfam`                    | 0.50 GB | interproscan | — |
+| `dfam_curated`            | 2.00 GB | repeatmasker, earlgrey | — |
+| `uniprot_sprot`           | 0.25 GB | braker3 | — |
+| `eggnog`                  | 8.50 GB | eggnog_mapper | — |
+| `kraken2_standard_8gb`    | 7.50 GB | kraken2, bracken | — |
+| `bowtie2_grch38_noalt`    | 3.50 GB | bowtie2 | hg38/bowtie2_index |
+| `dbsnp_grch38`            | 1.60 GB | gatk4 | hg38/dbsnp |
+| `mills_indels_grch38`     | 0.02 GB | gatk4 | hg38/known_indels |
+| `encode_blacklist_grch38` | 0.001 GB | macs3, tobias | hg38/blacklist |
+| `gencode_grch38`          | 0.05 GB | star, salmon, subread | hg38/ensembl_gtf |
+
+`bioflow db manifest` emits a [refgenie](https://refgenie.databio.org/)-compatible
+JSON mapping `<genome>/<asset>` → catalogued DB, so a lab already on
+refgenie can see which existing assets satisfy a bioflow requirement.
 
 ---
 

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/bioflow/__init__.py

@@ -1,6 +1,6 @@
 """bioflow - bioinformatics pipeline platform."""
 
-__version__ = "0.2.0"
+__version__ = "0.2.1"
 
 # Tier-A SDK — @stage / @pipeline / runtime config
 from bioflow.sdk import (  # noqa: E402,F401

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/bioflow/cli/__init__.py

@@ -18,6 +18,7 @@ Subcommands (MVP scope):
   bioflow update        Run the monthly registry update workflow.
   bioflow ncbi          Search / download genomes & proteins from NCBI.
   bioflow recipe        List or run a curated end-to-end recipe.
+  bioflow provenance    Inspect a recipe run's recorded provenance.
   bioflow llm           Opt-in LLM companion (explain / diagnose / …).
   bioflow setup         First-time LLM backend wizard.
 """
@@ -33,6 +34,7 @@ from bioflow.cli import (  # noqa: F401,E402
     llm,
     ncbi,
     pipelines,
+    provenance,
     recipe,
     setup,
     update,

{bioflowkit-0.2.0 → bioflowkit-0.2.1}/bioflow/cli/db.py

@@ -12,7 +12,7 @@ from bioflow.cli._app import app, console
 
 @app.command("db")
 def db_cmd(
-    action: str = typer.Argument(..., help="fetch | list | verify"),
+    action: str = typer.Argument(..., help="fetch | list | verify | manifest"),
     name: Optional[str] = typer.Argument(None, help="Database key (see 'bioflow db list')."),
     dest: Path = typer.Option(
         Path("data/references"),
@@ -21,23 +21,43 @@ def db_cmd(
     ),
     force: bool = typer.Option(False, "--force", "-f", help="Re-download even if file exists."),
 ) -> None:
-    """Fetch or verify reference databases (eggNOG / Pfam / Dfam / BUSCO / UniProt)."""
-    from bioflow.core.db import fetch_db, list_dbs, verify_db  # noqa: PLC0415
+    """Fetch / verify / list reference databases, or emit a refgenie manifest.
+
+    \b
+    list      Show every catalogued DB, its size, and which tools use it.
+    fetch     Download a DB into --dest (resumable, size-checked).
+    verify    Confirm a fetched DB is present (+ MD5 when registered).
+    manifest  Emit a refgenie-compatible asset manifest (JSON) mapping
+              genome/asset → catalogued DB, so labs on refgenie can see
+              which existing assets satisfy a bioflow requirement.
+    """
+    import json  # noqa: PLC0415
+
+    from bioflow.core.db import (  # noqa: PLC0415
+        fetch_db, list_dbs, refgenie_manifest, verify_db,
+    )
 
     if action == "list":
         rows = list_dbs()
         console.print("\n[bold]Available reference databases:[/]\n")
         for r in rows:
             used = ", ".join(r["used_by"])
+            tag = f"  [dim]({r['genome']}/{r['asset']})[/]" if r.get("genome") else ""
             console.print(
-                f"  [cyan]{r['key']:<20}[/]  {r['size_gb']:5.1f} GB  "
-                f"used by: {used}"
+                f"  [cyan]{r['key']:<24}[/]  {r['size_gb']:5.2f} GB  "
+                f"used by: {used}{tag}"
             )
             console.print(f"    {r['name']}")
             if r["notes"]:
                 console.print(f"    [dim]{r['notes']}[/]")
         return
 
+    if action == "manifest":
+        import sys  # noqa: PLC0415
+        manifest = refgenie_manifest(dest_root=dest)
+        sys.stdout.write(json.dumps(manifest, indent=2) + "\n")
+        return
+
     if not name:
         rprint("[red]Error:[/] 'name' argument required for fetch/verify.")
         raise typer.Exit(code=1)