biology/subread: Read alignment, quantification and mutation discovery

The Subread package comprises a suite of software programs for processing
next-gen sequencing read data including:

    Subread: a general-purpose read aligner
    Subjunc: a read aligner developed for aligning RNA-seq reads
    featureCounts: a software program developed for counting reads to genomic
    features such as genes, exons, promoters and genomic bins.
    Sublong: a long-read aligner that is designed based on seed-and-vote.
    exactSNP: discovers SNPs by testing signals against local background noise.

17 files changed, 366 insertions, 0 deletions

diff --git a/biology/subread/Makefile b/biology/subread/Makefile
new file mode 100644
index 000000000000..422b0c00b80a
--- /dev/null
+++ b/biology/subread/Makefile
@@ -0,0 +1,23 @@
+# $FreeBSD$
+
+PORTNAME=		subread
+DISTVERSION=		2.0.1
+DISTVERSIONSUFFIX=	-source
+CATEGORIES=		biology
+MASTER_SITES=		SF/${PORTNAME}/${PORTNAME}-${PORTVERSION}
+
+MAINTAINER=	jwb@FreeBSD.org
+COMMENT=	High-performance read alignment, quantification and mutation discovery
+
+LICENSE=	GPLv3
+LICENSE_FILE=	${WRKDIR}/${DISTNAME}/LICENSE
+
+LIB_DEPENDS=	libsysinfo.so:devel/libsysinfo
+
+USES=		gmake localbase:ldflags
+
+MAKEFILE=	${FILESDIR}/Makefile.pkg
+WRKSRC_SUBDIR=	src
+INSTALL_TARGET=	install-strip
+
+.include <bsd.port.mk>
diff --git a/biology/subread/distinfo b/biology/subread/distinfo
new file mode 100644
index 000000000000..b0a68a9a2d64
--- /dev/null
+++ b/biology/subread/distinfo
@@ -0,0 +1,3 @@
+TIMESTAMP = 1591305316
+SHA256 (subread-2.0.1-source.tar.gz) = d808eb5b1823c572cb45a97c95a3c5acb3d8e29aa47ec74e3ca1eb345787c17b
+SIZE (subread-2.0.1-source.tar.gz) = 23260381
diff --git a/biology/subread/files/Makefile.pkg b/biology/subread/files/Makefile.pkg
new file mode 100644
index 000000000000..2ae783e69902
--- /dev/null
+++ b/biology/subread/files/Makefile.pkg
@@ -0,0 +1,104 @@
+
+
+include makefile.version
+
+CCFLAGS = -DMAKE_FOR_EXON  -DMAKE_STANDALONE \
+	-DSUBREAD_VERSION=\"${SUBREAD_VERSION}\"  -D_FILE_OFFSET_BITS=64
+LDFLAGS += -DMAKE_FOR_EXON -DMAKE_STANDALONE -pthread -lsysinfo -lz -lm
+CC ?= gcc
+CC += ${CCFLAGS}
+
+BINS	= detectionCall sublong repair txUnique featureCounts removeDup \
+	exactSNP subread-buildindex subindel  subread-align subjunc \
+	qualityScores subread-fullscan propmapped flattenGTF \
+
+MKDIR	?= mkdir -p
+INSTALL	?= install
+DESTDIR	?= .
+PREFIX	?= local
+STRIP	?= strip
+
+ALL_LIBS= core core-junction core-indel sambam-file sublog gene-algorithms hashtable input-files sorted-hashtable gene-value-index exon-algorithms HelperFunctions interval_merge long-hashtable core-bigtable seek-zlib input-blc
+ALL_OBJECTS=$(addsuffix .o, ${ALL_LIBS})
+ALL_H=$(addsuffix .h, ${ALL_LIBS})
+ALL_C=$(addsuffix .c, ${ALL_LIBS})
+
+all: genRandomReads detectionCall sublong repair txUnique featureCounts removeDup exactSNP subread-buildindex subindel  subread-align subjunc qualityScores subread-fullscan propmapped flattenGTF # samMappedBases mergeVCF  testZlib
+	mkdir -p ../bin/utilities
+	mv longread-one/LRM longread-one/sublong
+	mv longread-one/sublong subread-align subjunc featureCounts  subindel exactSNP subread-buildindex ../bin/
+	mv detectionCall genRandomReads repair propmapped qualityScores removeDup subread-fullscan txUnique flattenGTF ../bin/utilities
+
+sublong: longread-one/longread-mapping.c  ${ALL_OBJECTS}
+	echo " " > longread-one/make.version
+	rm -f longread-one/*.o
+	cd longread-one && $(MAKE)
+
+genRandomReads: gen_rand_reads.c ${ALL_OBJECTS}
+	${CC} -o genRandomReads gen_rand_reads.c ${ALL_OBJECTS} ${LDFLAGS}
+
+flattenGTF: flattenAnnotations.c ${ALL_OBJECTS}
+	${CC} -o flattenGTF flattenAnnotations.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+detectionCall: detection-calls.c	 ${ALL_OBJECTS}
+	${CC} -o detectionCall detection-calls.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+repair: read-repair.c	 ${ALL_OBJECTS}
+	${CC} -o repair read-repair.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+txUnique: tx-unique.c tx-unique.h	 ${ALL_OBJECTS}
+	${CC} -o txUnique tx-unique.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+globalReassembly: global-reassembly.c ${ALL_OBJECTS}
+	${CC} -o globalReassembly  global-reassembly.c ${ALL_OBJECTS} ${LDFLAGS}
+
+propmapped: propmapped.c ${ALL_OBJECTS}
+	${CC} -o propmapped propmapped.c ${ALL_OBJECTS} ${LDFLAGS}
+
+exactSNP: SNPCalling.c SNPCalling.h  ${ALL_OBJECTS}
+	${CC}  -o exactSNP SNPCalling.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+subread-buildindex: index-builder.c subread.h ${ALL_OBJECTS}
+	${CC}  -o subread-buildindex  index-builder.c ${ALL_OBJECTS} ${LDFLAGS}
+
+removeDup: removeDupReads.c removeDupReads.h subread.h ${ALL_OBJECTS}
+	${CC} -o  removeDup removeDupReads.c  ${ALL_OBJECTS} ${LDFLAGS}
+
+subindel: SUBindel.c core.h subread.h  ${ALL_OBJECTS}
+	${CC} -o subindel SUBindel.c ${ALL_OBJECTS} ${LDFLAGS}
+
+featureCounts: readSummary.c subread.h  ${ALL_OBJECTS}
+	${CC} -o featureCounts readSummary.c ${ALL_OBJECTS} ${LDFLAGS}
+
+subread-align: core-interface-aligner.c  ${ALL_OBJECTS}
+	${CC} -o subread-align core-interface-aligner.c  ${ALL_OBJECTS} ${LDFLAGS} 
+
+subjunc: core-interface-subjunc.c  ${ALL_OBJECTS}
+	${CC} -o subjunc core-interface-subjunc.c  ${ALL_OBJECTS} ${LDFLAGS} 
+
+subtools: subtools.c ${ALL_OBJECTS}
+	${CC} -o subtools subtools.c ${ALL_OBJECTS} ${LDFLAGS} 
+
+qualityScores: qualityScores.c ${ALL_OBJECTS}
+	${CC} -o qualityScores qualityScores.c  ${ALL_OBJECTS}  ${LDFLAGS}
+
+subread-fullscan: fullscan.c  ${ALL_OBJECTS}
+	${CC} -o subread-fullscan fullscan.c  ${ALL_OBJECTS}  ${LDFLAGS}
+
+.PHONY:	clean install install-strip
+
+install-strip: install
+	${STRIP} ${DESTDIR}${PREFIX}/bin/*
+
+install:
+	${MKDIR} ${DESTDIR}${PREFIX}/bin
+	${INSTALL} -c \
+		../bin/exactSNP \
+		../bin/featureCounts \
+		../bin/sub* \
+		../bin/utilities/* \
+		${DESTDIR}${PREFIX}/bin
+
+clean:
+	rm -f core featureCounts exactSNP removeDup subread-buildindex \
+		${ALL_OBJECTS}
diff --git a/biology/subread/files/patch-HelperFunctions.c b/biology/subread/files/patch-HelperFunctions.c
new file mode 100644
index 000000000000..e10e44273b9f
--- /dev/null
+++ b/biology/subread/files/patch-HelperFunctions.c
@@ -0,0 +1,20 @@
+--- HelperFunctions.c.orig	2020-06-04 23:58:40 UTC
++++ HelperFunctions.c
+@@ -845,7 +845,7 @@ int strcmp_number(char * s1, char * s2)
+ 
+ int mac_str(char * str_buff)
+ {
+-#if defined(FREEBSD) || defined(__MINGW32__)
++#if defined(__FreeBSD__) || defined(__MINGW32__)
+ 	return 1;
+ #else
+ #ifdef MACOS
+@@ -2703,7 +2703,7 @@ void main(){
+ 
+ int get_free_total_mem(size_t * total, size_t * free_mem){
+ 
+-#ifdef FREEBSD
++#ifdef __FreeBSD__
+     return -1;
+ #endif
+ 
diff --git a/biology/subread/files/patch-core-junction.c b/biology/subread/files/patch-core-junction.c
new file mode 100644
index 000000000000..44567984a111
--- /dev/null
+++ b/biology/subread/files/patch-core-junction.c
@@ -0,0 +1,10 @@
+--- core-junction.c.orig	2018-11-08 15:19:02 UTC
++++ core-junction.c
+@@ -22,6 +22,7 @@
+ #include <string.h>
+ #include <ctype.h>
+ #include <assert.h>
++#include <unistd.h>
+ #include "subread.h"
+ #include "sublog.h"
+ #include "gene-value-index.h"
diff --git a/biology/subread/files/patch-gene-algorithms.c b/biology/subread/files/patch-gene-algorithms.c
new file mode 100644
index 000000000000..803acbd28909
--- /dev/null
+++ b/biology/subread/files/patch-gene-algorithms.c
@@ -0,0 +1,20 @@
+--- gene-algorithms.c.orig	2020-06-04 23:57:31 UTC
++++ gene-algorithms.c
+@@ -28,7 +28,7 @@
+ #include <sys/stat.h>
+ #include <sys/types.h>
+ 
+-#ifndef FREEBSD
++#ifndef __FreeBSD__
+ #include <sys/timeb.h>
+ #endif
+ 
+@@ -1563,7 +1563,7 @@ int load_offsets(gene_offset_t* offsets , const char i
+ 
+ double miltime(){
+ 	double ret;
+-	#ifdef FREEBSD
++	#ifdef __FreeBSD__
+ 	struct timeval tp;
+ 	struct timezone tz;
+ 	tz.tz_minuteswest=0;
diff --git a/biology/subread/files/patch-input-files.c b/biology/subread/files/patch-input-files.c
new file mode 100644
index 000000000000..37eca1d9e74d
--- /dev/null
+++ b/biology/subread/files/patch-input-files.c
@@ -0,0 +1,11 @@
+--- input-files.c.orig	2020-06-05 00:33:03 UTC
++++ input-files.c
+@@ -48,7 +48,7 @@ FILE * f_subr_open(const char * fname, const char * mo
+ #ifdef __MINGW32__
+ 		return fopen64(fname, mode);
+ #else
+-#if defined(__LP64__) || defined(_LP64) || defined(MACOS) 
++#if defined(__LP64__) || defined(_LP64) || defined(MACOS) || defined(__FreeBSD__)
+ 		return fopen(fname, mode);
+ #else
+ 		return fopen64(fname, mode);
diff --git a/biology/subread/files/patch-long-hashtable.c b/biology/subread/files/patch-long-hashtable.c
new file mode 100644
index 000000000000..cfb9557fa621
--- /dev/null
+++ b/biology/subread/files/patch-long-hashtable.c
@@ -0,0 +1,11 @@
+--- long-hashtable.c.orig	2020-06-04 23:59:33 UTC
++++ long-hashtable.c
+@@ -22,7 +22,7 @@
+ #include <string.h>
+ 
+ #ifndef MACOS
+-#ifndef FREEBSD
++#ifndef __FreeBSD__
+ #include <malloc.h>
+ #endif
+ #endif
diff --git a/biology/subread/files/patch-longread-one_LRMsorted-hashtable.c b/biology/subread/files/patch-longread-one_LRMsorted-hashtable.c
new file mode 100644
index 000000000000..f5cfe3d3cb12
--- /dev/null
+++ b/biology/subread/files/patch-longread-one_LRMsorted-hashtable.c
@@ -0,0 +1,15 @@
+--- longread-one/LRMsorted-hashtable.c.orig	2018-11-08 14:47:43 UTC
++++ longread-one/LRMsorted-hashtable.c
+@@ -22,11 +22,9 @@
+ #include <string.h>
+ #include "LRMsorted-hashtable.h"
+ 
+-#ifndef MACOS
+-#ifndef FREEBSD
++#if !defined(__APPLE__) && !defined(__FreeBSD__)
+ #include <malloc.h>
+ #endif
+-#endif
+ 
+ #include<math.h>
+ #include "LRMfile-io.h"
diff --git a/biology/subread/files/patch-longread-one_Makefile b/biology/subread/files/patch-longread-one_Makefile
new file mode 100644
index 000000000000..ef09f35cd441
--- /dev/null
+++ b/biology/subread/files/patch-longread-one_Makefile
@@ -0,0 +1,30 @@
+--- longread-one/Makefile.orig	2019-09-04 04:22:49 UTC
++++ longread-one/Makefile
+@@ -1,12 +1,12 @@
+-CC_EXEC = gcc
+ OPT_LEVEL = 3
+ 
+ include ../makefile.version
+ include make.version
+ 
+-CCFLAGS = -mtune=core2 ${MACOS} -O${OPT_LEVEL} -Wall  -DMAKE_FOR_EXON  -D MAKE_STANDALONE  -D_FILE_OFFSET_BITS=64 -DSUBREAD_VERSION=\"${SUBREAD_VERSION}\" ${WARNING_LEVEL} ${MINGW32}
+-LDFLAGS =  -lpthread -lz -lm -O${OPT_LEVEL} -DMAKE_FOR_EXON -D MAKE_STANDALONE # -DREPORT_ALL_THE_BEST
+-CC = ${CC_EXEC}  ${CCFLAGS}  -fmessage-length=0  -ggdb
++CC ?= gcc
++CFLAGS += -DMAKE_FOR_EXON  -D MAKE_STANDALONE  -D_FILE_OFFSET_BITS=64 \
++	-DSUBREAD_VERSION=\"${SUBREAD_VERSION}\" ${WARNING_LEVEL} ${MINGW32}
++LDFLAGS +=  -lpthread -lz -lm -O${OPT_LEVEL} -DMAKE_FOR_EXON -D MAKE_STANDALONE # -DREPORT_ALL_THE_BEST
+ 
+ ALL_LIBS=LRMsorted-hashtable LRMbase-index LRMchro-event LRMhelper LRMseek-zlib LRMfile-io LRMhashtable
+ ALL_OBJECTS=$(addsuffix .o, ${ALL_LIBS})
+@@ -20,8 +20,8 @@ clean:
+ 	rm -f *.o LRM
+ 
+ LRM: longread-mapping.c  ${ALL_OBJECTS} 
+-	${CC} -o LRM longread-mapping.c   ${ALL_OBJECTS}  ${LDFLAGS}
++	${CC} -o LRM ${CFLAGS} longread-mapping.c   ${ALL_OBJECTS}  ${LDFLAGS}
+ 
+ $(ALL_OBJECTS): $(ALL_C) $(ALL_H)
+-	$(CC) -o $@ -c $(subst .o,.c,$@)
++	$(CC) -o $@ -c ${CFLAGS} $(subst .o,.c,$@)
+ 	
diff --git a/biology/subread/files/patch-longread-one_longread-mapping.c b/biology/subread/files/patch-longread-one_longread-mapping.c
new file mode 100644
index 000000000000..d1304ef7d4d8
--- /dev/null
+++ b/biology/subread/files/patch-longread-one_longread-mapping.c
@@ -0,0 +1,37 @@
+--- longread-one/longread-mapping.c.orig	2019-09-04 04:22:49 UTC
++++ longread-one/longread-mapping.c
+@@ -30,7 +30,9 @@
+ #ifndef __MINGW32__
+ #include <sys/resource.h>
+ #endif
++#ifndef __FreeBSD__	// Deprecated on FreeBSD
+ #include <sys/timeb.h>
++#endif
+ #include <sys/stat.h>
+ #include <locale.h>
+ #include <ctype.h>
+@@ -223,9 +225,24 @@ int LRMvalidate_and_init_context(LRMcontext_t ** conte
+ 
+ double LRMmiltime(){
+ 	double ret;
++
++/* Why not use gettimeofday() on all platforms? */
++#ifdef __FreeBSD__
++
++	struct timeval tp;
++	struct timezone tz;
++	tz.tz_minuteswest=0;
++	tz.tz_dsttime=0;
++	gettimeofday(&tp,&tz);
++	ret = tp.tv_sec+ 0.001*0.001* tp.tv_usec; 
++
++#else
++
+ 	struct timeb trp;
+ 	ftime(&trp);
+ 	ret = trp.time*1.0+(trp.millitm*1.0/1000.0);
++
++#endif
+ 	return ret;
+ }
+ 
diff --git a/biology/subread/files/patch-read-repair.c b/biology/subread/files/patch-read-repair.c
new file mode 100644
index 000000000000..aebf9c944967
--- /dev/null
+++ b/biology/subread/files/patch-read-repair.c
@@ -0,0 +1,10 @@
+--- read-repair.c.orig	2018-11-08 15:20:06 UTC
++++ read-repair.c
+@@ -3,6 +3,7 @@
+ #include <string.h>
+ #include <getopt.h>
+ #include <zlib.h>
++#include <unistd.h>
+ #include "subread.h"
+ #include "core.h"
+ #include "input-files.h"
diff --git a/biology/subread/files/patch-sambam-file.c b/biology/subread/files/patch-sambam-file.c
new file mode 100644
index 000000000000..bb17ba9dc5c3
--- /dev/null
+++ b/biology/subread/files/patch-sambam-file.c
@@ -0,0 +1,10 @@
+--- sambam-file.c.orig	2018-11-08 15:20:35 UTC
++++ sambam-file.c
+@@ -30,6 +30,7 @@
+ #include <string.h>
+ #include <assert.h>
+ #include <ctype.h>
++#include <unistd.h>
+ #include "subread.h"
+ #include "core.h"
+ #include "gene-algorithms.h"
diff --git a/biology/subread/files/patch-sorted-hashtable.c b/biology/subread/files/patch-sorted-hashtable.c
new file mode 100644
index 000000000000..9a3637e9420a
--- /dev/null
+++ b/biology/subread/files/patch-sorted-hashtable.c
@@ -0,0 +1,15 @@
+--- sorted-hashtable.c.orig	2018-11-08 03:33:50 UTC
++++ sorted-hashtable.c
+@@ -23,11 +23,9 @@
+ #include <stdlib.h>
+ #include <string.h>
+ 
+-#ifndef MACOS
+-#ifndef FREEBSD
++#if !defined(__FreeBSD__) && !defined(__APPLE__)
+ #include <malloc.h>
+ #endif
+-#endif
+ 
+ #include<math.h>
+ #include"core.h"
diff --git a/biology/subread/files/patch-subread.h b/biology/subread/files/patch-subread.h
new file mode 100644
index 000000000000..3916daa6835f
--- /dev/null
+++ b/biology/subread/files/patch-subread.h
@@ -0,0 +1,11 @@
+--- subread.h.orig	2018-11-08 14:46:06 UTC
++++ subread.h
+@@ -102,7 +102,7 @@
+ 
+ #define USE_POSIX_MUTEX_LOCK
+ 
+-#if defined(MACOS) || defined(FREEBSD) || defined(USE_POSIX_MUTEX_LOCK)
++#if defined(__APPLE__) || defined(__FreeBSD__) || defined(USE_POSIX_MUTEX_LOCK)
+ typedef pthread_mutex_t subread_lock_t;
+ #define pthread_spinlock_t pthread_mutex_t
+ #define pthread_spin_lock pthread_mutex_lock
diff --git a/biology/subread/pkg-descr b/biology/subread/pkg-descr
new file mode 100644
index 000000000000..cb32c24836d6
--- /dev/null
+++ b/biology/subread/pkg-descr
@@ -0,0 +1,20 @@
+The Subread package comprises a suite of software programs for processing
+next-gen sequencing read data including:
+
+    Subread: a general-purpose read aligner which can align both genomic
+    DNA-seq and RNA-seq reads. It can also be used to discover genomic
+    mutations including short indels and structural variants.
+
+    Subjunc: a read aligner developed for aligning RNA-seq reads and for the
+    detection of exon-exon junctions. Gene fusion events can be detected as
+    well.
+
+    featureCounts: a software program developed for counting reads to genomic
+    features such as genes, exons, promoters and genomic bins.
+
+    Sublong: a long-read aligner that is designed based on seed-and-vote.
+
+    exactSNP: a SNP caller that discovers SNPs by testing signals against local
+    background noises.
+
+WWW: http://bioinf.wehi.edu.au/subread-package/
diff --git a/biology/subread/pkg-plist b/biology/subread/pkg-plist
new file mode 100644
index 000000000000..37404a072ab1
--- /dev/null
+++ b/biology/subread/pkg-plist
@@ -0,0 +1,16 @@
+bin/detectionCall
+bin/exactSNP
+bin/featureCounts
+bin/flattenGTF
+bin/genRandomReads
+bin/propmapped
+bin/qualityScores
+bin/removeDup
+bin/repair
+bin/subindel
+bin/subjunc
+bin/sublong
+bin/subread-align
+bin/subread-buildindex
+bin/subread-fullscan
+bin/txUnique