{bio}[GCC/11.3.0] CD-HIT v4.8.1, GenomeTools v1.6.2, Kent_tools v442, ...

easybuild/easyconfigs/c/CD-HIT/CD-HIT-4.8.1-GCC-11.3.0.eb

@@ -0,0 +1,42 @@
+# This file is an EasyBuild reciPY as per https://github.com/easybuilders/easybuild
+
+easyblock = 'MakeCp'
+
+name = 'CD-HIT'
+version = '4.8.1'
+
+homepage = 'https://github.com/weizhongli/cdhit/wiki'
+description = """ CD-HIT is a very widely used program for clustering and 
+ comparing protein or nucleotide sequences."""
+
+toolchain = {'name': 'GCC', 'version': '11.3.0'}
+toolchainopts = {'openmp': True}
+
+source_urls = ['https://github.com/weizhongli/cdhit/releases/download/V%(version)s/']
+sources = ['%(namelower)s-v%(version)s-2019-0228.tar.gz']
+checksums = ['26172dba3040d1ae5c73ff0ac6c3be8c8e60cc49fc7379e434cdf9cb1e7415de']
+
+dependencies = [
+    ('Perl', '5.34.1'),
+    ('zlib', '1.2.12'),
+]
+
+# make sure compilation flags are passed down (e.g. to enable OpenMP)
+buildopts = ' CC="$CXX" CCFLAGS="$CPPFLAGS $CXXFLAGS"'
+
+# put here the list of generated executables when compiling
+local_list_of_executables = ['cd-hit', 'cd-hit-est', 'cd-hit-2d', 'cd-hit-est-2d', 'cd-hit-div', 'cd-hit-454']
+
+# this is the real EasyBuild line to copy all the executables and perl scripts to "bin"
+files_to_copy = [(local_list_of_executables, 'bin'), (['*.pl'], 'bin'), 'README', 'doc', 'license.txt']
+
+fix_perl_shebang_for = ['bin/*.pl']
+
+sanity_check_paths = {
+    'files': ['bin/%s' % x for x in local_list_of_executables],
+    'dirs': [],
+}
+
+sanity_check_commands = ["cd-hit -h | grep 'CD-HIT version %(version)s'"]
+
+moduleclass = 'bio'

easybuild/easyconfigs/g/GenomeTools/GenomeTools-1.6.2-GCC-11.3.0.eb

@@ -0,0 +1,38 @@
+easyblock = 'ConfigureMake'
+
+name = 'GenomeTools'
+version = '1.6.2'
+
+homepage = 'http://genometools.org'
+description = "A comprehensive software library for efficient processing of structured genome annotations."
+
+toolchain = {'name': 'GCC', 'version': '11.3.0'}
+
+github_account = 'genometools'
+source_urls = [GITHUB_LOWER_SOURCE]
+sources = ['v%(version)s.tar.gz']
+checksums = ['974825ddc42602bdce3d5fbe2b6e2726e7a35e81b532a0dc236f6e375d18adac']
+
+# GenomeTools-1.6.2 has the following libraries bundled with it:
+# bzip2-1.0.6, cgilua-5.1.3, expat-2.0.1, lpeg-0.10.2, lua-5.1.5, luafilesystem-1.5.0, md5-1.2,
+# samtools-0.1.18, sqlite-3.33.0, tre-0.8.0, zlib-1.2.8
+#
+# Bundled libraries can be globally disabled with the option useshared=yes
+# However, it is preferable to use the bundled libraries due to the very old versions of some of them
+
+builddependencies = [('pkgconf', '1.8.0')]
+dependencies = [('Pango', '1.50.7')]
+
+skipsteps = ['configure']
+
+buildopts = 'useshared=no errorcheck=no cairo=yes threads=yes'
+installopts = 'prefix=%(installdir)s'
+
+sanity_check_paths = {
+    'files': ['bin/gt', 'bin/genometools-config', 'lib/libgenometools.a', 'lib/libgenometools.%s' % SHLIB_EXT],
+    'dirs': ['include', 'share'],
+}
+
+sanity_check_commands = ['gt -help']
+
+moduleclass = 'bio'

easybuild/easyconfigs/k/Kent_tools/Kent_tools-442-GCC-11.3.0.eb

@@ -0,0 +1,44 @@
+easyblock = 'MakeCp'
+
+name = 'Kent_tools'
+version = '442'
+
+homepage = 'https://genome.cse.ucsc.edu/'
+description = """Kent utilities: collection of tools used by the UCSC genome browser."""
+
+toolchain = {'name': 'GCC', 'version': '11.3.0'}
+source_urls = [
+    'https://hgdownload.cse.ucsc.edu/admin/exe/',
+    'https://hgdownload.cse.ucsc.edu/admin/exe/userApps.archive/',
+]
+sources = ['userApps.v%(version)s.src.tgz']
+patches = ['%(name)s-442_openssl.patch']
+checksums = [
+    {'userApps.v442.src.tgz': 'c54711b4a4c35de3c8c3094ed21b95a8b330e4132e19ef4c21020b24406d910c'},
+    {'Kent_tools-442_openssl.patch': '219ff93006553098c21d5ff05248163c952a5086e2547b2b29f88a2d1a0a2e53'},
+]
+
+dependencies = [
+    ('MariaDB', '10.9.3'),
+    ('libpng', '1.6.37'),
+    ('zlib', '1.2.12'),
+    ('util-linux', '2.38'),
+    ('OpenSSL', '1.1', '', SYSTEM),
+    ('freetype', '2.12.1'),
+]
+
+buildopts = 'CC="$CC" COPT="$CFLAGS -fcommon" PNGLIB="-L$EBROOTLIBPNG/lib -lpng" ZLIB="-L$EBROOTZLIB/lib -lz" '
+buildopts += 'SSL_DIR="$EBROOTOPENSSL" SSLDIR="$EBROOTOPENSSL" MYSQLLIBS="-L$EBROOTMARIADB/lib -lmariadb -lstdc++" '
+
+local_binaries = ['blat', 'bedPartition', 'getRna', 'liftOver', 'mafGene', 'splitFile', 'twoBitToFa']
+
+files_to_copy = [(['bin/*'], 'bin'), 'licenseBlat.txt', 'licenseUcscGenomeBrowser.txt']
+
+sanity_check_paths = {
+    'files': ['bin/%s' % x for x in local_binaries],
+    'dirs': [],
+}
+
+sanity_check_commands = ["%s 2>&1 | grep '^usage:'" % x for x in local_binaries]
+
+moduleclass = 'bio'

easybuild/easyconfigs/k/Kent_tools/Kent_tools-442_openssl.patch

@@ -0,0 +1,60 @@
+Patch to ensure SSL will be picked up by EasyBuild and not the OS. Important for older OpenSSL versions
+original Author: J. Sassmannshausen / GSTT
+updated for v442 by Samuel Moors (Vrije Universiteit Brussel)
+diff -ur userApps.orig/kent/src/inc/common.mk userApps/kent/src/inc/common.mk
+--- userApps.orig/kent/src/inc/common.mk	2023-01-18 01:19:26.000000000 +0100
++++ userApps/kent/src/inc/common.mk	2023-02-01 14:06:02.404382429 +0100
+@@ -133,7 +133,7 @@
+ ifneq (${SSL_DIR}, "/usr/include/openssl")
+   ifneq ($(UNAME_S),Darwin)
+     ifneq ($(wildcard ${SSL_DIR}),)
+-      L+=-L${SSL_DIR}/lib
++      L+=-L${SSL_DIR}/lib -lssl -lcrypto
+     endif
+   endif
+     HG_INC+=-I${SSL_DIR}/include
+@@ -142,44 +142,6 @@
+ ifeq (${IS_HGWDEV},yes)
+    L+=/hive/groups/browser/freetype/freetype-2.10.0/objs/.libs/libfreetype.a -lbz2
+    L+=/usr/lib64/libssl.a /usr/lib64/libcrypto.a -lkrb5 -lk5crypto -ldl
+-else
+-   ifeq (${CONDA_BUILD},1)
+-       L+=${PREFIX}/lib/libssl.a ${PREFIX}/lib/libcrypto.a -ldl
+-   else
+-     ifneq ($(wildcard /opt/homebrew/Cellar/openssl@3/3.0.7/lib/libssl.a),)
+-         L+=/opt/homebrew/Cellar/openssl@3/3.0.7/lib/libssl.a
+-     else
+-       ifneq ($(wildcard /opt/local/lib/libssl.a),)
+-         L+=/opt/local/lib/libssl.a
+-       else
+-         ifneq ($(wildcard /usr/lib/x86_64-linux-gnu/libssl.a),)
+-	   L+=/usr/lib/x86_64-linux-gnu/libssl.a
+-         else
+-           ifneq ($(wildcard /usr/local/opt/openssl/lib/libssl.a),)
+-              L+=/usr/local/opt/openssl/lib/libssl.a
+-           else
+-              L+=-lssl
+-           endif
+-         endif
+-       endif
+-     endif
+-     ifneq ($(wildcard /opt/homebrew/Cellar/openssl@3/3.0.7/lib/libcrypto.a),)
+-         L+=/opt/homebrew/Cellar/openssl@3/3.0.7/lib/libcrypto.a
+-     else
+-       ifneq ($(wildcard /opt/local/lib/libcrypto.a),)
+-          L+=/opt/local/lib/libcrypto.a
+-       else
+-          ifneq ($(wildcard /usr/local/opt/openssl/lib/libcrypto.a),)
+-             L+=/usr/local/opt/openssl/lib/libcrypto.a
+-          else
+-             L+=-lcrypto
+-          endif
+-       endif
+-     endif
+-     ifneq ($(wildcard /opt/homebrew/Cellar/zstd/1.5.2/lib/libzstd.a),)
+-          L+=/opt/homebrew/Cellar/zstd/1.5.2/lib/libzstd.a
+-     endif
+-   endif
+ endif
+
+ # autodetect where libm is installed

easybuild/easyconfigs/l/LTR_retriever/LTR_retriever-2.9.0-foss-2022a.eb

@@ -0,0 +1,39 @@
+# Author: Jasper Grimm (UoY)
+
+easyblock = 'Tarball'
+
+name = 'LTR_retriever'
+version = '2.9.0'
+
+homepage = 'https://github.com/oushujun/LTR_retriever'
+description = """LTR_retriever is a highly accurate and sensitive program for
+ identification of LTR retrotransposons; The LTR Assembly Index (LAI) is also
+ included in this package.
+"""
+
+toolchain = {'name': 'foss', 'version': '2022a'}
+
+github_account = 'oushujun'
+source_urls = [GITHUB_SOURCE]
+sources = ['v%(version)s.tar.gz']
+checksums = ['e2d94f6179c33990a77fa9fdcefb842c8481b4c30833c9c12cbbe54cb3fdda73']
+
+dependencies = [
+    ('Perl', '5.34.1'),
+    ('BLAST+', '2.13.0'),
+    ('RepeatMasker', '4.1.4'),
+    ('HMMER', '3.3.2'),
+    ('CD-HIT', '4.8.1'),
+    ('TRF', '4.09.1'),
+]
+
+sanity_check_paths = {
+    'files': ['LTR_retriever'],
+    'dirs': ['bin', 'database'],
+}
+
+sanity_check_commands = [("LTR_retriever -h 2>&1 | grep 'Usage: LTR_retriever'", '')]
+
+modextrapaths = {'PATH': ''}
+
+moduleclass = 'bio'

easybuild/easyconfigs/r/RECON/RECON-1.08-GCC-11.3.0.eb

@@ -0,0 +1,40 @@
+easyblock = 'MakeCp'
+
+name = 'RECON'
+version = '1.08'
+
+homepage = 'https://www.repeatmasker.org/RepeatModeler/'
+description = """Patched version of RECON to be used with RepeatModeler."""
+
+toolchain = {'name': 'GCC', 'version': '11.3.0'}
+
+source_urls = ['https://www.repeatmasker.org/RepeatModeler/']
+sources = ['%(name)s-%(version)s.tar.gz']
+checksums = ['699765fa49d18dbfac9f7a82ecd054464b468cb7521abe9c2bd8caccf08ee7d8']
+
+dependencies = [('Perl', '5.34.1')]
+
+start_dir = 'src'
+
+buildopts = 'CC="$CC"'
+
+local_bins = ['edgeredef', 'eledef', 'eleredef', 'famdef', 'imagespread']
+local_scripts = ['scripts/%s' % x for x in ['MSPCollect.pl', 'recon.pl']]
+
+files_to_copy = [
+    (['src/%s' % x for x in local_bins], 'bin'),
+    'scripts', 'COPYRIGHT', 'LICENSE', '00README',
+]
+
+postinstallcmds = ["sed -i 's|$path = \"\"|$path = \"%(installdir)s/bin\"|g' %(installdir)s/scripts/recon.pl"]
+
+sanity_check_paths = {
+    'files': ['bin/%s' % x for x in local_bins] + local_scripts,
+    'dirs': [],
+}
+
+sanity_check_commands = [("recon.pl 2>&1 | grep 'usage: recon seq_name_list_file MSP_file integer'", '')]
+
+modextrapaths = {'PATH': 'scripts'}
+
+moduleclass = 'bio'

easybuild/easyconfigs/r/RepeatModeler/RepeatModeler-2.0.4-foss-2022a.eb

@@ -0,0 +1,67 @@
+easyblock = 'Tarball'
+
+name = 'RepeatModeler'
+version = '2.0.4'
+
+homepage = 'https://www.repeatmasker.org/'
+description = """RepeatModeler is a de novo transposable element (TE) family identification and modeling package. """
+
+toolchain = {'name': 'foss', 'version': '2022a'}
+
+source_urls = ['https://www.repeatmasker.org/RepeatModeler']
+sources = [SOURCE_TAR_GZ]
+checksums = ['94aad46cc70911d48de3001836fc3165adb95b2b282b5c53ab0d1da98c27a6b6']
+
+dependencies = [
+    ('Perl', '5.34.1'),
+    ('TRF', '4.09.1'),
+    ('RMBlast', '2.13.0'),
+    ('RepeatMasker', '4.1.4'),
+    ('RepeatScout', '1.0.6'),
+    ('RECON', '1.08'),
+    ('Kent_tools', '442'),
+    ('CD-HIT', '4.8.1'),
+    # Optional (required for LTR pipeline):
+    ('MAFFT', '7.505', '-with-extensions'),
+    ('GenomeTools', '1.6.2'),   # for LTRHarvest
+    ('LTR_retriever', '2.9.0'),
+    ('TWL-NINJA', '0.98-cluster_only'),
+]
+
+_tmpl = "\"/'%s' =>/,/'value' => '/s|'value' =>.*|'value' => '%s'|\""
+
+postinstallcmds = [
+    ' '.join([
+        'sed -i',
+        '-e', _tmpl % ('CDHIT_DIR', '$EBROOTCDMINHIT/bin'),
+        '-e', _tmpl % ('GENOMETOOLS_DIR', '$EBROOTGENOMETOOLS/bin'),
+        '-e', _tmpl % ('LTR_RETRIEVER_DIR', '$EBROOTLTR_RETRIEVER'),
+        '-e', _tmpl % ('MAFFT_DIR', '$EBROOTMAFFT/bin'),
+        '-e', _tmpl % ('NINJA_DIR', '$EBROOTTWLMINNINJA/bin'),
+        '-e', _tmpl % ('RECON_DIR', '$EBROOTRECON/bin'),
+        '-e', _tmpl % ('REPEATMASKER_DIR', '$EBROOTREPEATMASKER'),
+        '-e', _tmpl % ('RMBLAST_DIR', '$EBROOTRMBLAST/bin'),
+        '-e', _tmpl % ('RSCOUT_DIR', '$EBROOTREPEATSCOUT'),
+        '-e', _tmpl % ('TRF_DIR', '$EBROOTTRF/bin'),
+        '-e', _tmpl % ('UCSCTOOLS_DIR', '$EBROOTKENT_TOOLS/bin'),
+        '%(installdir)s/RepModelConfig.pm',
+    ]),
+]
+
+_bins = ['BuildDatabase', 'LTRPipeline', 'RepeatClassifier', 'Refiner', 'TRFMask', 'util/CntSubst', 'util/Linup']
+
+fix_perl_shebang_for = _bins + ['util/*.pl']
+
+modextrapaths = {
+    'PATH': '',
+    'PERL5LIB': '',
+}
+
+sanity_check_paths = {
+    'files': _bins + ['Job.pm', 'RepeatUtil.pm', 'RepModelConfig.pm'],
+    'dirs': ['Libraries', 'Matrices', 'util'],
+}
+
+sanity_check_commands = ["RepeatModeler -help | grep 'RepeatModeler - Model repetitive DNA'"]
+
+moduleclass = 'bio'

easybuild/easyconfigs/r/RepeatScout/RepeatScout-1.0.6-GCC-11.3.0.eb

@@ -0,0 +1,38 @@
+# Author: Jasper Grimm (UoY)
+
+easyblock = 'ConfigureMake'
+
+name = 'RepeatScout'
+version = '1.0.6'
+
+homepage = 'https://www.repeatmasker.org/'
+description = """De Novo Repeat Finder, Price A.L., Jones N.C. and Pevzner P.A. Developed and
+ tested with our multiple sequence version of RepeatScout ( 1.0.6 )"""
+
+toolchain = {'name': 'GCC', 'version': '11.3.0'}
+
+source_urls = ['https://www.repeatmasker.org']
+sources = ['RepeatScout-%(version)s.tar.gz']
+checksums = ['31a44cf648d78356aec585ee5d3baf936d01eaba43aed382d9ac2d764e55b716']
+
+dependencies = [
+    ('Perl', '5.34.1'),
+    ('TRF', '4.09.1'),
+]
+
+skipsteps = ['configure']
+
+prebuildopts = "sed -i -e 's|mkdir|mkdir -p|g' -e 's|VERSION = .*|VERSION = %(version)s|g' Makefile &&"
+
+installopts = 'INSTDIR="%(installdir)s"'
+
+sanity_check_paths = {
+    'files': [name, 'build_lmer_table', 'README'],
+    'dirs': [],
+}
+
+sanity_check_commands = [("RepeatScout 2>&1 | grep 'Usage:\nRepeatScout'", '')]
+
+modextrapaths = {'PATH': ''}
+
+moduleclass = 'bio'