Integrate Hackathon vrs-in-a-box into vrs-python

vrs-in-a-box/.gitignore

@@ -0,0 +1,2 @@
+seqrepo-*
+GCF_*

vrs-in-a-box/Dockerfile

@@ -0,0 +1,57 @@
+# Builder image
+FROM python:3.12-slim AS build
+
+# Either 'GRCh38' or 'GRCh37'
+ARG ASSEMBLY="GRCh38"
+ARG VRS_PYTHON_VERSION="2.1.2"
+
+# Install packages needed for the build
+RUN apt-get update && apt-get upgrade -y && apt-get install -y \
+    curl \
+    git \
+    libpq-dev \
+    python3-pip \
+    python3-venv \
+    rsync \
+    zlib1g-dev \
+    postgresql \
+    unzip \
+    ;
+
+WORKDIR /vrs-python
+
+# Setup the virtual env for vrs-python
+RUN python3 -m venv /vrs-python/venv
+ENV PATH=/vrs-python/venv/bin:$PATH
+
+# Install vrs-python
+RUN /vrs-python/venv/bin/python3 -m pip install -U setuptools
+RUN /vrs-python/venv/bin/python3 -m pip install ga4gh.vrs[extras]==${VRS_PYTHON_VERSION}
+
+# Download and unpack seqrepo files
+COPY /seqrepo-${ASSEMBLY}.zip /
+RUN unzip /seqrepo-${ASSEMBLY}.zip -d /
+
+# Final image
+FROM python:3.12-slim AS vrs-python
+ARG ASSEMBLY
+ENV ASSEMBLY=${ASSEMBLY}
+
+# Install runtime required packages
+RUN apt-get update && apt-get install -y libpq-dev
+
+# Copy over artifacts from the builder
+COPY --from=build /vrs-python /vrs-python
+COPY --from=build /seqrepo-${ASSEMBLY} /seqrepo-${ASSEMBLY}
+
+# Copy over run script
+COPY ./run.sh /run.sh
+
+# Set environment variables
+ENV GA4GH_VRS_DATAPROXY_URI="seqrepo+file:///seqrepo-${ASSEMBLY}/master"
+ENV VIRTUAL_ENV=/vrs-python/venv
+ENV PATH=/vrs-python/venv/bin:$PATH
+
+WORKDIR /
+
+ENTRYPOINT [ "/run.sh" ]

vrs-in-a-box/README.md

@@ -0,0 +1,57 @@
+# VRS-in-a-Box VCF Annotator for Assembly GRCh37 or GRCh38
+VRS-in-a-Box is a single Docker image that is able to annotate a VCF with VRS IDs
+using the `vrs-annotate` tool that is part of `vrs-python`.  The Docker image includes
+all the dependencies required for VRS ID computation of genomic variants on one of the
+assembled chromosomes for a specific reference assembly.
+
+The Docker image is kept to a minimum by creating an instance of `seqrepo` that
+contains only the assembled chromosomes for a single reference assembly.
+
+Prebuilt images are available in Docker Hub: https://hub.docker.com/u/ga4gh
+
+## Building VRS-in-a-Box
+The following instructions describe how to build a VRS-in-a-Box image from scratch.
+
+
+#### Create the Build Environment
+Install any [prerequisites](https://github.com/biocommons/biocommons.seqrepo#requirements)
+needed for `seqrepo` and create a Python virtual environment.
+```bash
+python -m venv venv
+source venv/bin/activate
+pip install biocommons.seqrepo
+```
+
+#### Download the Reference Assembly Sequences
+```bash
+# GRCh38
+curl -O https://ftp.ncbi.nlm.nih.gov/genomes/refseq/vertebrate_mammalian/Homo_sapiens/all_assembly_versions/GCF_000001405.26_GRCh38/GCF_000001405.26_GRCh38_genomic.fna.gz
+# GRCh37
+curl -O https://ftp.ncbi.nlm.nih.gov/genomes/refseq/vertebrate_mammalian/Homo_sapiens/all_assembly_versions/GCF_000001405.13_GRCh37/GCF_000001405.13_GRCh37_genomic.fna.gz
+```
+
+#### Build Seqrepo
+```bash
+# GRCh38
+bash seqrepo.sh GCF_000001405.26_GRCh38_genomic.fna.gz GRCh38
+# GRCh37
+bash seqrepo.sh GCF_000001405.26_GRCh37_genomic.fna.gz GRCh37
+```
+
+## Build Images for Each Assembly
+```shell
+# GRCh38
+docker build --build-arg ASSEMBLY=GRCh38 --build-arg VRS_PYTHON_VERSION=2.1.2 -t ga4gh/vrs-vcf-annotator-grch38:2.1.2 .
+# GRCh37
+docker build --build-arg ASSEMBLY=GRCh37 --build-arg VRS_PYTHON_VERSION=2.1.2 -t ga4gh/vrs-vcf-annotator-grch37:2.1.2 .
+```
+
+## Running the Image in Docker
+Run the image to annotate the VCF file `NA12878.vcf` in the current directory:
+```shell
+docker run -it --rm -v $(pwd):/input ga4gh/vrs-vcf-annotator-grch38:2.1.2 /input/NA12878.vcf --vcf-out /input/NA12878_with_vrs.vcf
+```
+Run the image to annotate the VCF file `NA12878.vcf` in the current directory and capture the VRS objects in a separate file:
+```shell
+docker run -it --rm -v $(pwd):/input ga4gh/vrs-vcf-annotator-grch38:2.1.2 /input/NA12878.vcf --vcf-out /input/NA12878_with_vrs.vcf --ndjson-out /input/vrs-objects.json
+```

vrs-in-a-box/run.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+/vrs-python/venv/bin/vrs-annotate vcf --assembly ${ASSEMBLY} $@

vrs-in-a-box/seqrepo.sh

@@ -0,0 +1,69 @@
+#!/bin/bash
+set -e  # Exit on any error
+
+
+# Check if an argument is provided
+if [ "$#" -ne 2 ]; then
+    echo "Usage: $0 <fasta_file> <assembly_name>"
+    exit 1
+fi
+
+FASTA_FILE=$1
+ASSEMBLY_NAME=$2
+
+export SEQREPO_ROOT=seqrepo-$ASSEMBLY_NAME
+
+# Check if the file exists
+if [ ! -f "$FASTA_FILE" ]; then
+    echo "ERROR: File '$FASTA_FILE' not found!" 1>&2
+    exit 1
+fi
+
+# Check if openrsync is in the output of /usr/bin/rsync --version
+RSYNC_EXE=${RSYNC:-rsync}
+if $RSYNC_EXE --version | grep -q "openrsync"; then
+    if [ -x "/opt/homebrew/bin/rsync" ]; then
+        RSYNC_OPTION="--rsync-exe /opt/homebrew/bin/rsync"
+    else
+        echo "ERROR: seqrepo requires 'rsync' (not 'openrsync') to run." 1>&2
+        echo "       On MacOS rsync can be installed with Homebrew." 1>&2
+        echo "       Set the RSYNC environment variable to specify the rsync executable to use." 1>&2
+        exit 1
+    fi
+else
+    RSYNC_OPTION="--rsync-exe $RSYNC_EXE"
+fi
+
+# Initialize seqrepo
+echo "Initializing seqrepo in $SEQREPO_ROOT..."
+seqrepo -r seqrepo --root-directory $SEQREPO_ROOT $RSYNC_OPTION init
+
+# Load the provided FASTA file
+echo "Loading FASTA file: $FASTA_FILE"
+seqrepo -r seqrepo --root-directory $SEQREPO_ROOT $RSYNC_OPTION load "$FASTA_FILE" -n NCBI
+
+# Add assembly names
+echo "Adding assembly names..."
+seqrepo -r seqrepo --root-directory $SEQREPO_ROOT $RSYNC_OPTION add-assembly-names
+
+echo "Initial seqrepo build completed"
+
+# vrs-annotate checks that GRCh38:1 is in the database as a sanity check
+#   so for non-GRCh38, we have to add a dummy record so it will run
+if [ "$ASSEMBLY_NAME" != "GRCh38" ]; then
+sqlite3 $SEQREPO_ROOT/master/aliases.sqlite3 <<EOF
+    INSERT INTO seqalias (seq_id, namespace, alias, added, is_current)
+    VALUES ('Ya6Rs7DHhDeg7YaOSg1EoNi3U_nQ9SvO', 'GRCh38', '1', DATE('now'), 1);
+EOF
+sqlite3 $SEQREPO_ROOT/master/sequences/db.sqlite3 <<EOF
+INSERT INTO seqinfo (seq_id, len, alpha, added, relpath)
+VALUES ('Ya6Rs7DHhDeg7YaOSg1EoNi3U_nQ9SvO', 248956422, 'ACGMNRT', date('now'), '2025/0331/1411/1743430287.165611.fa.bgz');
+EOF
+echo "Spiked in GRCh38:1 sequence alias to make vrs-annotate run"
+fi
+# Create a tar.gz archive of the seqrepo directory
+echo "Creating seqrepo archive..."
+zip -r "${SEQREPO_ROOT}.zip" ${SEQREPO_ROOT}
+rm -fR $SEQREPO_ROOT
+
+echo "Archive created ${SEQREPO_ROOT}.zip"