biology/snpeff: Genetic variant annotation and effect prediction toolbox

Genetic variant annotation and functional effect prediction toolbox. It
annotates and predicts the effects of genetic variants on genes and proteins
(such as amino acid changes).

7 files changed, 180 insertions, 0 deletions

diff --git a/biology/Makefile b/biology/Makefile
index 40af21229bfa..7c6bfbddf1d1 100644
--- a/biology/Makefile
+++ b/biology/Makefile
@@ -168,6 +168,7 @@
     SUBDIR += sim4
     SUBDIR += slclust
     SUBDIR += smithwaterman
+    SUBDIR += snpeff
     SUBDIR += spoa
     SUBDIR += stacks
     SUBDIR += star
diff --git a/biology/snpeff/Makefile b/biology/snpeff/Makefile
new file mode 100644
index 000000000000..3973e9a9a792
--- /dev/null
+++ b/biology/snpeff/Makefile
@@ -0,0 +1,41 @@
+PORTNAME=	snpeff
+DISTVERSION=	5.0
+CATEGORIES=	biology java python
+MASTER_SITES=	https://snpeff.blob.core.windows.net/versions/
+DISTNAME=	snpEff_latest_core
+
+MAINTAINER=	jwb@FreeBSD.org
+COMMENT=	Genetic variant annotation and effect prediction toolbox
+
+LICENSE=	MIT
+LICENSE_FILE=	${WRKSRC}/LICENSE.md
+
+RUN_DEPENDS=	bash:shells/bash
+
+USES=		python:3.6+ shebangfix zip
+USE_JAVA=	yes
+
+JAVA_VERSION=	12+
+SHEBANG_FILES=	scripts/* scripts/gsa/*
+
+NO_BUILD=	yes
+NO_ARCH=	yes
+WRKSRC=		${WRKDIR}/snpEff
+
+LIBEXEC_DIR=	${PREFIX}/libexec/snpeff
+
+do-install:
+	${MKDIR} ${STAGEDIR}${JAVAJARDIR}/snpeff
+	${MKDIR} ${STAGEDIR}${LIBEXEC_DIR}
+	${INSTALL_SCRIPT} ${WRKSRC}/scripts/snpEff ${STAGEDIR}${PREFIX}/bin
+	${INSTALL_SCRIPT} ${WRKSRC}/scripts/snpSift ${STAGEDIR}${PREFIX}/bin
+	${INSTALL_DATA} ${WRKSRC}/*.jar ${STAGEDIR}${JAVAJARDIR}/snpeff
+	${INSTALL_DATA} ${WRKSRC}/*.config ${STAGEDIR}${JAVAJARDIR}/snpeff
+	cd ${WRKSRC}/scripts && ${COPYTREE_BIN} . ${STAGEDIR}${LIBEXEC_DIR}
+
+pre-configure:
+	${REINPLACE_CMD} -e 's|%%JAVAJARDIR%%|${JAVAJARDIR}|g' \
+			 -e 's|%%PREFIX%%|${PREFIX}|g' \
+			 ${WRKSRC}/scripts/snpEff ${WRKSRC}/scripts/snpSift
+
+.include <bsd.port.mk>
diff --git a/biology/snpeff/distinfo b/biology/snpeff/distinfo
new file mode 100644
index 000000000000..895315e4a8c3
--- /dev/null
+++ b/biology/snpeff/distinfo
@@ -0,0 +1,3 @@
+TIMESTAMP = 1618353537
+SHA256 (snpEff_latest_core.zip) = 85d907b5dd9e9008a0cf245956e3c9077a31e45f21a1b580d9b98a53fd8dcb9d
+SIZE (snpEff_latest_core.zip) = 45828841
diff --git a/biology/snpeff/files/patch-scripts_snpEff b/biology/snpeff/files/patch-scripts_snpEff
new file mode 100644
index 000000000000..dace484960d7
--- /dev/null
+++ b/biology/snpeff/files/patch-scripts_snpEff
@@ -0,0 +1,27 @@
+--- scripts/snpEff.orig	2021-04-13 22:55:13 UTC
++++ scripts/snpEff
+@@ -13,9 +13,10 @@
+ # Created by: Brad Chapman
+ #---------------------------------------------------------------------------------------------------
+ 
+-jardir="$(cd "$(dirname "$0")" && cd ".." && pwd -P)"
++jardir=%%JAVAJARDIR%%/snpeff
++userdir=~/snpEff
++JAVA_HOME=%%PREFIX%%/openjdk12
+ 
+-java=java
+ if [ -e "$JAVA_HOME/bin/java" ]
+ then
+ 	java="$JAVA_HOME/bin/java"
+@@ -43,8 +44,9 @@ if [ "$jvm_mem_opts" == "" ]; then
+     jvm_mem_opts="$default_jvm_mem_opts"
+ fi
+ if [[ "$pass_args" != "" && ! "$pass_args" =~ "-c " ]]; then
+-    pass_args="$pass_args -c ${jardir}/snpEff.config"
++    pass_args="$pass_args -c ${userdir}/snpEff.config"
+ fi
+ 
++mkdir -p ~/snpEff
++cp ${jardir}/*.config ~/snpEff
+ exec $java $jvm_mem_opts $jvm_prop_opts -jar ${jardir}/snpEff.jar $pass_args
+-exit
diff --git a/biology/snpeff/files/patch-scripts_snpSift b/biology/snpeff/files/patch-scripts_snpSift
new file mode 100644
index 000000000000..838875a344f8
--- /dev/null
+++ b/biology/snpeff/files/patch-scripts_snpSift
@@ -0,0 +1,19 @@
+--- scripts/snpSift.orig	2021-04-13 22:40:00 UTC
++++ scripts/snpSift
+@@ -13,9 +13,9 @@
+ # Created by: Brad Chapman
+ #---------------------------------------------------------------------------------------------------
+ 
+-jardir="$(cd "$(dirname "$0")" && cd ".." && pwd -P)"
++jardir=%%JAVAJARDIR%%/snpeff
++JAVA_HOME=%%PREFIX%%/openjdk12
+ 
+-java=java
+ if [ -e "$JAVA_HOME/bin/java" ]
+ then
+ 	java="$JAVA_HOME/bin/java"
+@@ -44,4 +44,3 @@ if [ "$jvm_mem_opts" == "" ]; then
+ fi
+ 
+ exec $java $jvm_mem_opts $jvm_prop_opts -jar ${jardir}/SnpSift.jar $pass_args
+-exit
diff --git a/biology/snpeff/pkg-descr b/biology/snpeff/pkg-descr
new file mode 100644
index 000000000000..cf47db97eb96
--- /dev/null
+++ b/biology/snpeff/pkg-descr
@@ -0,0 +1,14 @@
+Genetic variant annotation and functional effect prediction toolbox. It
+annotates and predicts the effects of genetic variants on genes and proteins
+(such as amino acid changes).
+
+Features:
+
+    Supports over 38,000 genomes
+    Standard ANN annotation format
+    Cancer variants analysis
+    GATK compatible (-o gatk)
+    HGVS notation
+    Sequence Ontology standardized terms
+
+WWW: http://snpeff.sourceforge.net/
diff --git a/biology/snpeff/pkg-plist b/biology/snpeff/pkg-plist
new file mode 100644
index 000000000000..a004332b0b41
--- /dev/null
+++ b/biology/snpeff/pkg-plist
@@ -0,0 +1,75 @@
+bin/snpEff
+bin/snpSift
+libexec/snpeff/1kg.sh
+libexec/snpeff/annotate_demo.sh
+libexec/snpeff/annotate_demo_GATK.sh
+libexec/snpeff/bedEffOnePerLine.pl
+libexec/snpeff/buildDbNcbi.sh
+libexec/snpeff/cgShore.pl
+libexec/snpeff/cgShore.sh
+libexec/snpeff/countColumns.py
+libexec/snpeff/db.pl
+libexec/snpeff/extractSequences.pl
+libexec/snpeff/fasta2tab.pl
+libexec/snpeff/fastaSample.pl
+libexec/snpeff/fastaSplit.pl
+libexec/snpeff/fastqSplit.pl
+libexec/snpeff/filterBy.py
+libexec/snpeff/gffRemovePhase.pl
+libexec/snpeff/gsa/bayesFactor_correction_scoreCount.r
+libexec/snpeff/gsa/bayesFactor_correction_scoreCount.sh
+libexec/snpeff/gsa/bayesFactor_correction_scoreCount_max10.sh
+libexec/snpeff/gsa/checkGeneNames.py
+libexec/snpeff/gsa/create_sets.bds
+libexec/snpeff/gsa/geneSetOverlap.py
+libexec/snpeff/gsa/geneSetOverlap.sort.txt
+libexec/snpeff/gsa/geneSetsGtex.py
+libexec/snpeff/gsa/pvalue_correction_scoreCount.r
+libexec/snpeff/gsa/pvalue_correction_scoreCount.sh
+libexec/snpeff/gsa/pvalue_correction_scoreCount_min10.sh
+libexec/snpeff/isutf8.py
+libexec/snpeff/join.pl
+libexec/snpeff/joinSnpEff.pl
+libexec/snpeff/make_and_commit_docs.sh
+libexec/snpeff/make_dbNSFP.sh
+libexec/snpeff/nextProt_filter.pl
+libexec/snpeff/ped2vcf.py
+libexec/snpeff/plot.pl
+libexec/snpeff/plotHistogram.pl
+libexec/snpeff/plotLabel.pl
+libexec/snpeff/plotMA.pl
+libexec/snpeff/plotQQ.pl
+libexec/snpeff/plotQQsubsample.pl
+libexec/snpeff/plotSmoothScatter.pl
+libexec/snpeff/plotXY.pl
+libexec/snpeff/queue.pl
+libexec/snpeff/sam2fastq.pl
+libexec/snpeff/sample.pl
+libexec/snpeff/snpEff
+libexec/snpeff/snpEff.bak
+libexec/snpeff/snpEff.orig
+libexec/snpeff/snpSift
+libexec/snpeff/snpSift.orig
+libexec/snpeff/snpSift_filter_sample_to_number.pl
+libexec/snpeff/sortLine.py
+libexec/snpeff/splitChr.pl
+libexec/snpeff/statsNum.pl
+libexec/snpeff/swapCols.pl
+libexec/snpeff/transpose.pl
+libexec/snpeff/txt2fa.pl
+libexec/snpeff/txt2vcf.py
+libexec/snpeff/uniqCount.pl
+libexec/snpeff/uniqCut.pl
+libexec/snpeff/vcfAnnFirst.py
+libexec/snpeff/vcfBareBones.pl
+libexec/snpeff/vcfEffHighest.ORI.py
+libexec/snpeff/vcfEffOnePerLine.pl
+libexec/snpeff/vcfFilterSamples.pl
+libexec/snpeff/vcfInfoOnePerLine.pl
+libexec/snpeff/vcfOnlyAlts.pl
+libexec/snpeff/vcfReduceGenotypes.pl
+libexec/snpeff/vcfRefCorrect.py
+libexec/snpeff/wigSplit.pl
+%%JAVAJARDIR%%/snpeff/SnpSift.jar
+%%JAVAJARDIR%%/snpeff/snpEff.config
+%%JAVAJARDIR%%/snpeff/snpEff.jar