Codex/add pacbio isoform tool containers

CHANGELOG.md

@@ -20,6 +20,17 @@
 
 ### Added
 
+- Latest stable containers for the PacBio full-length isoform sequencing
+  note's transcript-visualization and isoform-classification/quantification
+  tools: Swan 3.2, ggtranscript 1.0.0, SQANTI3 6.0.2, TALON 6.0.1, Cerberus
+  1.1, LAPA 0.0.5, FLAIR 3.0.1, lr-kallisto 0.52.0, IsoQuant 4.0.0, Bambu
+  3.14.0, and Oarfish 0.10.3. Each image includes a build-time functional
+  verification and a versioned GHCR publishing target.
+- Latest stable containers for all tools in the PacBio full-length isoform
+  sequencing note's differential-analysis sections: tappAS 1.1.3, DESeq2
+  1.52.0, DRIMSeq 1.40.0, DEXSeq 1.58.0, and SUPPA2 2.4. Each image includes a
+  build-time version and functional smoke test and is published under its
+  versioned GHCR tag.
 - HOMER 5.1 (`images/homer/5.1`) with bedtools 2.31.1 and Python 3.12 for
   FASTA-mode motif discovery, published as `ghcr.io/yeolab/homer:5.1`.
 - skipper-clipper-compare 1.0.0 (`images/skipper-clipper-compare/1.0.0`): the

README.md

@@ -43,6 +43,58 @@ write`, already declared in the workflow. If organization policy prevents a
 publish, allow GitHub Actions to create packages for this repository or replace
 the token with an approved package-write token.
 
+## PacBio full-length isoform analysis
+
+The tools listed under **Differential transcript analysis** and **Differential
+splicing usage** in PacBio's full-length isoform sequencing application note
+are available as individually versioned images:
+
+| Application-note section | Tool | Image |
+| --- | --- | --- |
+| Differential transcript analysis | tappAS 1.1.3 | `ghcr.io/yeolab/tappas:1.1.3` |
+| Differential transcript analysis | DESeq2 1.52.0 | `ghcr.io/yeolab/deseq2:1.52.0` |
+| Differential transcript analysis | DRIMSeq 1.40.0 | `ghcr.io/yeolab/drimseq:1.40.0` |
+| Differential splicing usage | DEXSeq 1.58.0 | `ghcr.io/yeolab/dexseq:1.58.0` |
+| Differential splicing usage | SUPPA2 2.4 | `ghcr.io/yeolab/suppa2:2.4` |
+
+DESeq2, DRIMSeq, and DEXSeq use the Bioconductor 3.23 release on R 4.6.
+SUPPA2 is a command-line image whose arguments are passed directly to
+`suppa.py`. tappAS is an x86-64 GUI image and needs an X11 display; see each
+image directory's README for launch examples and resource requirements. Every
+Dockerfile contains a build-time smoke test, so GitHub Actions publishes an
+image only after its installed version and a representative operation pass.
+
+The application note's transcript-visualization tools are also available:
+
+| Application-note section | Tool | Image |
+| --- | --- | --- |
+| Transcript visualization | Swan 3.2 | `ghcr.io/yeolab/swan:3.2` |
+| Transcript visualization | ggtranscript 1.0.0 | `ghcr.io/yeolab/ggtranscript:1.0.0` |
+
+The latest stable releases of the note's isoform classification and
+quantification tools are packaged as:
+
+| Tool | Image |
+| --- | --- |
+| SQANTI3 6.0.2 | `ghcr.io/yeolab/sqanti3:6.0.2` |
+| TALON 6.0.1 | `ghcr.io/yeolab/talon:6.0.1` |
+| Cerberus 1.1 | `ghcr.io/yeolab/cerberus:1.1` |
+| LAPA 0.0.5 | `ghcr.io/yeolab/lapa:0.0.5` |
+| FLAIR 3.0.1 | `ghcr.io/yeolab/flair:3.0.1` |
+| lr-kallisto / kallisto LongKmer 0.52.0 | `ghcr.io/yeolab/lr-kallisto:0.52.0` |
+| IsoQuant 4.0.0 | `ghcr.io/yeolab/isoquant:4.0.0` |
+| Bambu 3.14.0 | `ghcr.io/yeolab/bambu:3.14.0` |
+| Oarfish 0.10.3 | `ghcr.io/yeolab/oarfish:0.10.3` |
+
+Swan and the Python command-line tools use checksum-pinned release artifacts;
+ggtranscript is pinned to the current upstream commit because the project has
+not published GitHub releases. SQANTI3 and TALON use exact Bioconda builds,
+Bambu uses Bioconductor 3.23, and lr-kallisto uses kallisto's official
+LongKmer binary. The per-image READMEs document entrypoints, architecture, and
+upstream version caveats. Build-time tests exercise transcript parsing,
+annotation conversion, database creation, indexing, or quantification as
+appropriate instead of checking only `--version` output.
+
 ## Pulling a Singularity/Apptainer image from GHCR
 
 Install [Apptainer](https://apptainer.org/) (or SingularityCE), then pull a

images/SQANTI3/6.0.2/Dockerfile

@@ -0,0 +1,19 @@
+FROM quay.io/biocontainers/sqanti3:6.0.2--hdfd78af_1@sha256:3ebeb658fb2e0e308e73cba50dd3b97c63963d8225f303f298ad8ee76264939d
+
+LABEL software="SQANTI3" \
+      software.version="6.0.2" \
+      about.summary="Quality control and classification of long-read transcript isoforms" \
+      about.url="https://github.com/ConesaLab/SQANTI3" \
+      about.license="GPL-3.0-only" \
+      about.tags="Genomics,long-read-RNA,isoform-classification"
+
+ENV MPLBACKEND=Agg \
+    PYTHONPATH=/usr/local/share/sqanti3-6.0.2-1 \
+    PYTHONWARNINGS=ignore::DeprecationWarning
+
+COPY verify.py /tmp/verify.py
+RUN python /tmp/verify.py && rm /tmp/verify.py
+
+WORKDIR /work
+ENTRYPOINT []
+CMD ["bash"]

images/SQANTI3/6.0.2/README.md

@@ -0,0 +1,15 @@
+# SQANTI3 6.0.2
+
+This image is pinned to the official BioContainers image for SQANTI3's exact
+Bioconda 6.0.2 build and includes the complete aligner, command-line, Python,
+R, and report-generation stack.
+
+```bash
+docker run --rm -it -v "$PWD:/work" ghcr.io/yeolab/sqanti3:6.0.2
+```
+
+The build verifies both SQANTI3's reported version and R dependencies, checks
+the external tools, and parses a synthetic two-exon genePred transcript through
+SQANTI3's own parser.
+
+Upstream: <https://github.com/ConesaLab/SQANTI3>

images/SQANTI3/6.0.2/verify.py

@@ -0,0 +1,31 @@
+import subprocess
+import tempfile
+from pathlib import Path
+
+from src.config import __version__
+from src.parsers import isoforms_parser
+
+
+assert __version__ == "6.0.2"
+assert "SQANTI3 6.0.2" in subprocess.check_output(
+    ["sqanti3_qc.py", "--version"], text=True
+)
+
+Path("/test/logs").mkdir(parents=True, exist_ok=True)
+with tempfile.TemporaryDirectory() as directory:
+    gene_pred = Path(directory) / "isoforms.genePred"
+    gene_pred.write_text(
+        "tx1\tchr1\t+\t0\t200\t0\t200\t2\t0,100,\t50,200,\t0\tgene1\tcmpl\tcmpl\t0,0,\n"
+    )
+    parsed = isoforms_parser(str(gene_pred))
+    assert len(parsed["chr1"]) == 1
+    assert parsed["chr1"][0].junctions == [(50, 100)]
+
+for executable in ("gffread", "gmap", "minimap2", "samtools", "STAR"):
+    subprocess.run(["bash", "-c", f"command -v {executable}"], check=True)
+
+subprocess.run(
+    ["Rscript", "-e", "stopifnot(as.character(packageVersion('RColorConesa')) == '1.0.0')"],
+    check=True,
+)
+print("SQANTI3 6.0.2 smoke test passed")

images/bambu/3.14.0/Dockerfile

@@ -0,0 +1,16 @@
+FROM bioconductor/bioconductor_docker:RELEASE_3_23@sha256:821dbf9ac119eac41f177531c7ca8fc7084c99c04eb218a1aa7f52cb63bad5d9
+
+LABEL software="Bambu" \
+      software.version="3.14.0" \
+      about.summary="Long-read transcript discovery and quantification" \
+      about.url="https://bioconductor.org/packages/3.23/bioc/html/bambu.html" \
+      about.license="GPL-3.0-only" \
+      about.tags="Genomics,long-read-RNA,isoform-quantification"
+
+RUN Rscript -e 'BiocManager::install("bambu", ask = FALSE, update = FALSE)'
+
+COPY verify.R /tmp/verify.R
+RUN Rscript /tmp/verify.R && rm /tmp/verify.R
+
+WORKDIR /work
+CMD ["R"]

images/bambu/3.14.0/README.md

@@ -0,0 +1,12 @@
+# Bambu 3.14.0
+
+This image installs Bambu 3.14.0 from Bioconductor 3.23 on R 4.6.
+
+```bash
+docker run --rm -it -v "$PWD:/work" ghcr.io/yeolab/bambu:3.14.0
+```
+
+The build verifies the package version and converts a synthetic two-exon GTF
+into Bambu's prepared annotation representation.
+
+Upstream: <https://bioconductor.org/packages/3.23/bioc/html/bambu.html>

images/bambu/3.14.0/verify.R

@@ -0,0 +1,12 @@
+suppressPackageStartupMessages(library(bambu))
+stopifnot(as.character(packageVersion("bambu")) == "3.14.0")
+
+gtf <- tempfile(fileext = ".gtf")
+writeLines(c(
+  'chr1\tsmoke\ttranscript\t1\t200\t.\t+\t.\tgene_id "g1"; transcript_id "t1";',
+  'chr1\tsmoke\texon\t1\t50\t.\t+\t.\tgene_id "g1"; transcript_id "t1"; exon_id "e1";',
+  'chr1\tsmoke\texon\t101\t200\t.\t+\t.\tgene_id "g1"; transcript_id "t1"; exon_id "e2";'
+), gtf)
+annotations <- prepareAnnotations(gtf)
+stopifnot(length(annotations) == 1L)
+message("Bambu 3.14.0 smoke test passed")

images/cerberus/1.1/Dockerfile

@@ -0,0 +1,65 @@
+FROM python:3.10-slim@sha256:9d53d8d4c0e882f61913025db53b3aec4ef74336082a9ab47d8a14e9e8329b00
+
+LABEL software="Cerberus" \
+      software.version="1.1" \
+      about.summary="Transcript start, end, and intron-chain annotation" \
+      about.url="https://github.com/mortazavilab/cerberus" \
+      about.license="MIT" \
+      about.tags="Genomics,long-read-RNA,isoform-classification"
+
+ARG LAPA_URL="https://files.pythonhosted.org/packages/64/cc/e9059f9e9108f5a9a22e08164e29be3a937bab6c7636a8c2e51bf5b85ceb/lapa-0.0.5.tar.gz"
+ARG LAPA_SHA256="3240405cbdd0e68c344676bc780e62368359024726f0ef2c04ef487d49793669"
+ARG SWAN_URL="https://files.pythonhosted.org/packages/39/8e/153cc2311f3b7e3888540a18689673b312b3e72c68ef65cce9f7c482f8ed/swan_vis-3.2.tar.gz"
+ARG SWAN_SHA256="b7da08a52cc37592378428f72a22a7e994af8d63118c8949c2b758b5dfc7c1db"
+ARG CERBERUS_URL="https://github.com/mortazavilab/cerberus/archive/refs/tags/v1.1.tar.gz"
+ARG CERBERUS_SHA256="c1693ef8ab48cdf841e769eb49d5d957b1d426635a6f7fdfe162914673d7de9f"
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends build-essential ca-certificates curl patch \
+    && rm -rf /var/lib/apt/lists/* \
+    && curl -fsSL "$LAPA_URL" -o /tmp/lapa.tar.gz \
+    && echo "$LAPA_SHA256  /tmp/lapa.tar.gz" | sha256sum -c - \
+    && curl -fsSL "$SWAN_URL" -o /tmp/swan.tar.gz \
+    && echo "$SWAN_SHA256  /tmp/swan.tar.gz" | sha256sum -c - \
+    && pip install --no-cache-dir \
+        "numpy==1.23.0" \
+        "pandas==1.5.3" \
+        "scipy==1.10.1" \
+        "matplotlib==3.7.5" \
+        "h5py==3.8.0" \
+        "blosc2==2.0.0" \
+        "tables==3.8.0" \
+        "pandarallel==1.6.5" \
+        "anndata==0.9.2" \
+        "scanpy==1.9.8" \
+        "numba==0.57.1" \
+        "scikit-learn==1.3.2" \
+        "umap-learn==0.5.5" \
+        "sparse==0.14.0" \
+        "python-ternary==1.0.8" \
+        "cython<3" \
+    && pip install --no-cache-dir --no-build-isolation "sorted-nearest==0.0.33" \
+    && pip install --no-cache-dir "pyranges==0.1.4" \
+    && pip install --no-cache-dir /tmp/lapa.tar.gz \
+    && pip install --no-cache-dir /tmp/swan.tar.gz \
+    && rm -rf /tmp/lapa.tar.gz /tmp/swan.tar.gz \
+    && apt-get purge -y --auto-remove build-essential
+
+COPY pyranges-compat.patch /tmp/pyranges-compat.patch
+
+RUN curl -fsSL "$CERBERUS_URL" -o /tmp/cerberus.tar.gz \
+    && echo "$CERBERUS_SHA256  /tmp/cerberus.tar.gz" | sha256sum -c - \
+    && mkdir /tmp/cerberus \
+    && tar -xzf /tmp/cerberus.tar.gz --strip-components=1 -C /tmp/cerberus \
+    && sed -i "s/version='0.0.idk'/version='1.1'/" /tmp/cerberus/setup.py \
+    && patch -d /tmp/cerberus -p1 < /tmp/pyranges-compat.patch \
+    && pip install --no-cache-dir --no-deps /tmp/cerberus \
+    && rm -rf /tmp/cerberus.tar.gz /tmp/cerberus /tmp/pyranges-compat.patch \
+    && apt-get purge -y --auto-remove curl patch
+
+COPY verify.sh /tmp/verify.sh
+RUN bash /tmp/verify.sh && rm /tmp/verify.sh
+
+WORKDIR /work
+ENTRYPOINT ["cerberus"]
+CMD ["--help"]

images/cerberus/1.1/README.md

@@ -0,0 +1,16 @@
+# Cerberus 1.1
+
+This image installs the latest Cerberus tag from a checksum-pinned source
+archive. Upstream's placeholder package version is normalized to 1.1 so the
+installed metadata agrees with the immutable tag.
+The image also carries a compatibility patch that preserves strand
+metadata when Cerberus reconstructs a PyRanges object with PyRanges 0.1.4.
+
+```bash
+docker run --rm -v "$PWD:/work" ghcr.io/yeolab/cerberus:1.1 --help
+```
+
+Arguments are passed directly to `cerberus`. The build converts a synthetic
+two-exon GTF to both TSS BED and intron-chain TSV outputs.
+
+Upstream: <https://github.com/mortazavilab/cerberus>

images/cerberus/1.1/pyranges-compat.patch

@@ -0,0 +1,25 @@
+diff --git a/cerberus/cerberus.py b/cerberus/cerberus.py
+index 35a87f3..ad34177 100644
+--- a/cerberus/cerberus.py
++++ b/cerberus/cerberus.py
+@@ -232,0 +233 @@ def number_gtf_ends(bed, gtf, mode):
++    gr_gtf['Strand'] = gr_gtf['Strand'].astype(str)
+@@ -237,0 +239 @@ def number_gtf_ends(bed, gtf, mode):
++    bed['Strand'] = bed['Strand'].astype(str)
+@@ -245 +247 @@ def number_gtf_ends(bed, gtf, mode):
+-        temp = pr.PyRanges(gr_gtf.features.tss().df[cols])
++        temp = gr_gtf.features.tss().df[cols]
+@@ -247 +249,3 @@ def number_gtf_ends(bed, gtf, mode):
+-        temp = pr.PyRanges(gr_gtf.features.tes().df[cols])
++        temp = gr_gtf.features.tes().df[cols]
++    temp['Strand'] = temp['Strand'].astype(str)
++    temp = pr.PyRanges(temp)
+@@ -1665,0 +1670 @@ def get_ends_from_gtf(gtf, mode, dist, slack):
++        gr_gtf['Strand'] = gr_gtf['Strand'].astype(str)
+@@ -1678 +1683,3 @@ def get_ends_from_gtf(gtf, mode, dist, slack):
+-    bed = pr.PyRanges(bed.rename({'ThickStart': 'gene_id'}, axis=1))
++    bed = bed.rename({'ThickStart': 'gene_id'}, axis=1)
++    bed['Strand'] = bed['Strand'].astype(str)
++    bed = pr.PyRanges(bed)
+@@ -1778,0 +1786 @@ def get_ics_from_gtf(gtf):
++    df['Strand'] = df['Strand'].astype(str)

images/cerberus/1.1/verify.sh

@@ -0,0 +1,26 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+python - <<'PY'
+from importlib.metadata import version
+assert version("cerberus") == "1.1"
+import cerberus.cerberus
+PY
+python -m pip check
+cerberus --help >/dev/null
+
+work="$(mktemp -d)"
+trap 'rm -rf "$work"' EXIT
+printf '%b\n' \
+  'chr1\tsmoke\tgene\t1\t200\t.\t+\t.\tgene_id "g1"; gene_name "G1";' \
+  'chr1\tsmoke\ttranscript\t1\t200\t.\t+\t.\tgene_id "g1"; transcript_id "t1"; gene_name "G1"; transcript_name "T1";' \
+  'chr1\tsmoke\texon\t1\t50\t.\t+\t.\tgene_id "g1"; transcript_id "t1"; exon_number "1";' \
+  'chr1\tsmoke\texon\t101\t200\t.\t+\t.\tgene_id "g1"; transcript_id "t1"; exon_number "2";' \
+  > "$work/input.gtf"
+
+cerberus gtf_to_bed --gtf "$work/input.gtf" --mode tss -o "$work/tss.bed"
+cerberus gtf_to_ics --gtf "$work/input.gtf" -o "$work/ics.tsv"
+test -s "$work/tss.bed"
+test -s "$work/ics.tsv"
+grep -q 'g1_1' "$work/ics.tsv"
+echo "Cerberus 1.1 smoke test passed"

images/deseq2/1.52.0/Dockerfile

@@ -0,0 +1,15 @@
+FROM bioconductor/bioconductor_docker:RELEASE_3_23@sha256:821dbf9ac119eac41f177531c7ca8fc7084c99c04eb218a1aa7f52cb63bad5d9
+
+LABEL software="DESeq2" \
+      software.version="1.52.0" \
+      about.summary="Differential gene expression analysis based on the negative binomial distribution" \
+      about.url="https://bioconductor.org/packages/3.23/bioc/html/DESeq2.html" \
+      about.license="LGPL-3.0-or-later" \
+      about.tags="Genomics,RNA-seq,differential-expression"
+
+RUN Rscript -e 'BiocManager::install("DESeq2", ask = FALSE, update = FALSE)'
+
+COPY verify.R /tmp/verify.R
+RUN Rscript /tmp/verify.R && rm /tmp/verify.R
+
+CMD ["R"]

images/deseq2/1.52.0/README.md

@@ -0,0 +1,15 @@
+# DESeq2 1.52.0
+
+DESeq2 1.52.0 from Bioconductor 3.23 on R 4.6. The image build loads the
+package and runs a small two-condition differential-expression analysis.
+
+```bash
+docker run --rm -it -v "$PWD:/work" -w /work ghcr.io/yeolab/deseq2:1.52.0
+```
+
+The default command starts R. Run an analysis script directly with:
+
+```bash
+docker run --rm -v "$PWD:/work" -w /work \
+  ghcr.io/yeolab/deseq2:1.52.0 Rscript analysis.R
+```

images/deseq2/1.52.0/verify.R

@@ -0,0 +1,20 @@
+suppressPackageStartupMessages(library(DESeq2))
+
+stopifnot(packageVersion("DESeq2") == "1.52.0")
+
+set.seed(52)
+counts <- matrix(rnbinom(600, mu = 100, size = 2), nrow = 100)
+counts[1:10, 4:6] <- counts[1:10, 4:6] + 150L
+storage.mode(counts) <- "integer"
+col_data <- data.frame(
+  condition = factor(rep(c("control", "treated"), each = 3)),
+  row.names = paste0("sample", seq_len(ncol(counts)))
+)
+colnames(counts) <- row.names(col_data)
+
+dds <- DESeqDataSetFromMatrix(counts, col_data, design = ~ condition)
+dds <- DESeq(dds, quiet = TRUE)
+res <- results(dds)
+
+stopifnot(nrow(res) == nrow(counts), "log2FoldChange" %in% colnames(res))
+message("DESeq2 ", packageVersion("DESeq2"), " smoke test passed")

images/dexseq/1.58.0/Dockerfile

@@ -0,0 +1,15 @@
+FROM bioconductor/bioconductor_docker:RELEASE_3_23@sha256:821dbf9ac119eac41f177531c7ca8fc7084c99c04eb218a1aa7f52cb63bad5d9
+
+LABEL software="DEXSeq" \
+      software.version="1.58.0" \
+      about.summary="Inference of differential exon usage in RNA-seq" \
+      about.url="https://bioconductor.org/packages/3.23/bioc/html/DEXSeq.html" \
+      about.license="GPL-3.0-or-later" \
+      about.tags="Genomics,RNA-seq,differential-exon-usage"
+
+RUN Rscript -e 'BiocManager::install("DEXSeq", ask = FALSE, update = FALSE)'
+
+COPY verify.R /tmp/verify.R
+RUN Rscript /tmp/verify.R && rm /tmp/verify.R
+
+CMD ["R"]