biology/fastx-toolkit: CLI tools for FASTA/FASTQ files preprocessing

The FASTX-Toolkit is a collection of command line tools for
Short-Reads FASTA/FASTQ files preprocessing.

7 files changed, 109 insertions, 0 deletions

diff --git a/biology/fastx-toolkit/DESCR b/biology/fastx-toolkit/DESCR
new file mode 100644
index 00000000000..e3e97cad062
--- /dev/null
+++ b/biology/fastx-toolkit/DESCR
@@ -0,0 +1,2 @@
+The FASTX-Toolkit is a collection of command line tools for
+Short-Reads FASTA/FASTQ files preprocessing.
diff --git a/biology/fastx-toolkit/Makefile b/biology/fastx-toolkit/Makefile
new file mode 100644
index 00000000000..cba4d57e44a
--- /dev/null
+++ b/biology/fastx-toolkit/Makefile
@@ -0,0 +1,27 @@
+# $NetBSD: Makefile,v 1.1 2022/04/05 18:48:57 bacon Exp $
+
+DISTNAME=	fastx-toolkit-0.0.14.6
+CATEGORIES=	biology
+MASTER_SITES=	${MASTER_SITE_GITHUB:=agordon/}
+GITHUB_PROJECT=	fastx_toolkit
+GITHUB_TAG=	ea0ca83ba24dce80c20ca589b838a281fe5deb0c
+
+MAINTAINER=	bacon@NetBSD.org
+HOMEPAGE=	https://github.com/agordon/fastx_toolkit
+COMMENT=	CLI tools for Short-Reads FASTA/FASTQ files preprocessing
+LICENSE=	gnu-agpl-v3
+
+USE_LANGUAGES=	c c++
+USE_TOOLS+=	autoconf automake autoreconf bash pkg-config perl:run
+USE_LIBTOOL=	yes
+GNU_CONFIGURE=	yes
+REPLACE_PERL+=	scripts/*.pl
+REPLACE_BASH+=	scripts/fastq_quality_boxplot_graph.sh
+REPLACE_BASH+=	scripts/fastx_nucleotide_distribution_graph.sh
+REPLACE_BASH+=	scripts/fastx_nucleotide_distribution_line_graph.sh
+
+pre-configure:
+	${RUN} cd ${WRKSRC} && autoreconf -vif
+
+.include "../../devel/libgtextutils/buildlink3.mk"
+.include "../../mk/bsd.pkg.mk"
diff --git a/biology/fastx-toolkit/PLIST b/biology/fastx-toolkit/PLIST
new file mode 100644
index 00000000000..04e3aedde22
--- /dev/null
+++ b/biology/fastx-toolkit/PLIST
@@ -0,0 +1,21 @@
+@comment $NetBSD: PLIST,v 1.1 2022/04/05 18:48:57 bacon Exp $
+bin/fasta_clipping_histogram.pl
+bin/fasta_formatter
+bin/fasta_nucleotide_changer
+bin/fastq_masker
+bin/fastq_quality_boxplot_graph.sh
+bin/fastq_quality_converter
+bin/fastq_quality_filter
+bin/fastq_quality_trimmer
+bin/fastq_to_fasta
+bin/fastx_artifacts_filter
+bin/fastx_barcode_splitter.pl
+bin/fastx_clipper
+bin/fastx_collapser
+bin/fastx_nucleotide_distribution_graph.sh
+bin/fastx_nucleotide_distribution_line_graph.sh
+bin/fastx_quality_stats
+bin/fastx_renamer
+bin/fastx_reverse_complement
+bin/fastx_trimmer
+bin/fastx_uncollapser
diff --git a/biology/fastx-toolkit/distinfo b/biology/fastx-toolkit/distinfo
new file mode 100644
index 00000000000..ae2ae8a4960
--- /dev/null
+++ b/biology/fastx-toolkit/distinfo
@@ -0,0 +1,8 @@
+$NetBSD: distinfo,v 1.1 2022/04/05 18:48:57 bacon Exp $
+
+BLAKE2s (fastx-toolkit-0.0.14.6-ea0ca83ba24dce80c20ca589b838a281fe5deb0c.tar.gz) = 973b56d722a2796aa28a04de56ef9273ce5a99fae5cf40bdbf9ef796acb2b6ef
+SHA512 (fastx-toolkit-0.0.14.6-ea0ca83ba24dce80c20ca589b838a281fe5deb0c.tar.gz) = bb9cc9685a37e0a88b5110ad48856d4160dc1a29e6b76f8cdf78baf12a9cea97c35f711b1aef270bba5d44b42c8e6bcc31ef9eae239de785fe1bd38cd4994d3e
+Size (fastx-toolkit-0.0.14.6-ea0ca83ba24dce80c20ca589b838a281fe5deb0c.tar.gz) = 259875 bytes
+SHA1 (patch-m4_Makefile.am) = f8cb138860f0a3708a6690ea6cd33302b35bce76
+SHA1 (patch-src_fasta__formatter_fasta__formatter.cpp) = 5fc622320186c5415cbcc4ae97cfc24f68983949
+SHA1 (patch-src_libfastx_fastx.h) = 5a387ac575a5269bc03e3f09ab752889efb8870c
diff --git a/biology/fastx-toolkit/patches/patch-m4_Makefile.am b/biology/fastx-toolkit/patches/patch-m4_Makefile.am
new file mode 100644
index 00000000000..e85265c188a
--- /dev/null
+++ b/biology/fastx-toolkit/patches/patch-m4_Makefile.am
@@ -0,0 +1,14 @@
+$NetBSD: patch-m4_Makefile.am,v 1.1 2022/04/05 18:48:57 bacon Exp $
+
+Do not install unrelated m4 files.
+
+--- m4/Makefile.am.orig	2017-08-14 18:08:32.000000000 +0000
++++ m4/Makefile.am
+@@ -17,6 +17,6 @@ m4macros = ax_c_long_long.m4 \
+ 	   ax_cxx_header_stdcxx_tr1.m4
+ 
+ # The following is boilerplate
+-m4data_DATA = $(m4macros) 
++#m4data_DATA = $(m4macros) 
+ EXTRA_DIST = $(m4data_DATA) 
+ 
diff --git a/biology/fastx-toolkit/patches/patch-src_fasta__formatter_fasta__formatter.cpp b/biology/fastx-toolkit/patches/patch-src_fasta__formatter_fasta__formatter.cpp
new file mode 100644
index 00000000000..e2a5756115f
--- /dev/null
+++ b/biology/fastx-toolkit/patches/patch-src_fasta__formatter_fasta__formatter.cpp
@@ -0,0 +1,14 @@
+$NetBSD: patch-src_fasta__formatter_fasta__formatter.cpp,v 1.1 2022/04/05 18:48:57 bacon Exp $
+
+# Silence compile error
+
+--- src/fasta_formatter/fasta_formatter.cpp.orig	2022-04-03 12:45:51.295476843 +0000
++++ src/fasta_formatter/fasta_formatter.cpp
+@@ -103,6 +103,7 @@ void parse_command_line(int argc, char* 
+ 		switch(opt) {
+ 		case 'h':
+ 			usage();
++			break;
+ 		
+ 		case 'i':
+ 			input_filename = optarg;
diff --git a/biology/fastx-toolkit/patches/patch-src_libfastx_fastx.h b/biology/fastx-toolkit/patches/patch-src_libfastx_fastx.h
new file mode 100644
index 00000000000..0a25126f420
--- /dev/null
+++ b/biology/fastx-toolkit/patches/patch-src_libfastx_fastx.h
@@ -0,0 +1,23 @@
+$NetBSD: patch-src_libfastx_fastx.h,v 1.1 2022/04/05 18:48:57 bacon Exp $
+
+# Limit packing to this one structure to avoid compatibility issues
+
+--- src/libfastx/fastx.h.orig	2018-05-16 14:50:08 UTC
++++ src/libfastx/fastx.h
+@@ -58,7 +58,7 @@ typedef enum {
+ 	OUTPUT_SAME_AS_INPUT=3
+ } OUTPUT_FILE_TYPE;
+ 
+-#pragma pack(1) 
++#pragma pack(push,1) 
+ typedef struct 
+ {
+ 	/* Record data - common for FASTA/FASTQ */
+@@ -115,6 +115,7 @@ typedef struct 
+ 	FILE*	input;
+ 	FILE*	output;
+ } FASTX ;
++#pragma pack(pop)
+ 
+ 
+ void fastx_init_reader(FASTX *pFASTX, const char* filename,