{bio}[foss/2022a] FunGAP v1.1.1 w/ Python 3.10.4

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1-foss-2022a.eb

@@ -0,0 +1,105 @@
+easyblock = 'Tarball'
+
+name = 'FunGAP'
+version = '1.1.1'
+
+homepage = 'https://github.com/CompSynBioLab-KoreaUniv/FunGAP'
+description = "Fungal Genome Annotation Pipeline using evidence-based gene model evaluation."
+
+toolchain = {'name': 'foss', 'version': '2022a'}
+
+# Tag v1.1.1 points to an old commit still in v1.1.0
+# pull correct sources from specific commit
+_commit = '924f3ba080f98cbf181b0b21601e095619479ce6'
+
+source_urls = ['https://github.com/CompSynBioLab-KoreaUniv/FunGAP/archive']
+sources = [{'download_filename': '%s.tar.gz' % _commit, 'filename': SOURCE_TAR_GZ}]
+patches = [
+    'FunGAP-%(version)s_fix-snap-detection.patch',
+    'FunGAP-%(version)s_relax-dependency-checks.patch',
+    'FunGAP-%(version)s_replace-deprecated-pa-repeatmodeler.patch',
+    'FunGAP-%(version)s_fix-maker-exes.patch',
+    'FunGAP-%(version)s_fix-augustus-calls-in-runbreaker.patch',
+]
+checksums = [
+    {'FunGAP-1.1.1.tar.gz': '3d827c4b11452afdd51b71766e0e3193b7efad31db4536606115f2cac0b964c8'},
+    {'FunGAP-1.1.1_fix-snap-detection.patch': 'f782224ce186e1e2d8953898122b79d616f8a749b00ec662ec5a3fa1903550fa'},
+    {'FunGAP-1.1.1_relax-dependency-checks.patch': '73f9ae2a20cf03c34f852642a28d3b7e9858250f653ff1e711c3c9a56ae2fd77'},
+    {'FunGAP-1.1.1_replace-deprecated-pa-repeatmodeler.patch':
+     'c528aab74a070d6eedc9ff2097b6149e8d36759b9ff93c46302994179a38a774'},
+    {'FunGAP-1.1.1_fix-maker-exes.patch': 'de344cf45dad047ab46a17462e19ce0789a89071c1484cbdedd9d898f9601bd0'},
+    {'FunGAP-1.1.1_fix-augustus-calls-in-runbreaker.patch':
+     '150b6f79f95d0c8abece42231c5dcedf8610a811db9cc63ac6cc047f0dcfdc64'},
+]
+
+dependencies = [
+    ('Python', '3.10.4'),
+    ('Perl', '5.34.1'),
+    ('AUGUSTUS', '3.5.0'),
+    ('BamTools', '2.5.2'),
+    ('bcbio-gff', '0.7.0'),
+    ('BRAKER', '2.1.6'),
+    ('BUSCO', '5.4.5'),
+    ('HISAT2', '2.2.1'),
+    ('MAKER', '3.01.04'),
+    ('matplotlib', '3.5.2'),
+    ('networkx', '2.8.4'),
+    ('PfamScan', '1.6'),
+    ('RepeatModeler', '2.0.4'),
+    ('SAMtools', '1.16.1'),
+    ('SNAP-HMM', '20221022'),
+    ('Trinity', '2.15.1'),
+    ('wget', '1.21.3'),
+]
+
+exts_defaultclass = 'PythonPackage'
+exts_default_options = {
+    'source_urls': [PYPI_SOURCE],
+    'download_dep_fail': True,
+    'use_pip': True,
+}
+
+exts_list = [
+    ('markdown2', '2.4.8', {
+        'checksums': ['90475aca3d9c8e7df6d70c51de5bbbe9edf7fcf6a380bd1044d321500f5445da'],
+    }),
+]
+
+# FunGAP needs the Pfam database (280 MB download, 1.5 GB on disk)
+# if it's already available in your system change _pfam_db_dir to its location
+_pfam_db_dir = '%(installdir)s/db'
+_pfam_db_url = 'https://ftp.ebi.ac.uk/pub/databases/Pfam/current_release'
+_pfam_db_install_cmds = [
+    "mkdir -p %s" % _pfam_db_dir,
+    "wget -P %s %s/{Pfam-A.hmm.gz,Pfam-A.hmm.dat.gz,active_site.dat.gz}" % (_pfam_db_dir, _pfam_db_url),
+    "gzip -d %s/*.gz" % _pfam_db_dir,
+    "hmmpress %s/Pfam-A.hmm" % _pfam_db_dir,
+]
+
+postinstallcmds = [
+    # install Pfam DB if missing
+    "if [ ! -f %s/Pfam-A.hmm ]; then %s; fi" % (_pfam_db_dir, " && ".join(_pfam_db_install_cmds)),
+    # execute set_dependencies.py, which creates configuration file and runs some quick tests
+    ("cd %%(installdir)s && ./set_dependencies.py --pfam_db_path %s --genemark_path $EBROOTGENEMARKMINET "
+     "--maker_path $EBROOTMAKER/bin --snap_path $EBROOTSNAPMINHMM/bin" % _pfam_db_dir),
+]
+
+sanity_check_paths = {
+    'files': ['fungap.py'],
+    'dirs': ['lib/python%(pyshortver)s/site-packages']
+}
+
+modextrapaths = {
+    'PATH': '',
+    'PYTHONPATH': ['', 'lib/python%(pyshortver)s/site-packages']
+}
+
+modextravars = {
+    'FUNGAP_DIR': '%(installdir)s',
+}
+
+sanity_check_commands = [
+    ('fungap.py', '--help'),
+]
+
+moduleclass = 'bio'

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1_fix-augustus-calls-in-runbreaker.patch

@@ -0,0 +1,54 @@
+Set paths to AUGUSTUS installation preferably from its environment variables
+Fix location of getAnnoFastaFromJoingenes.py script from AUGUSTUS
+see https://github.com/CompSynBioLab-KoreaUniv/FunGAP/pull/98
+author: Alex Domingo (Vrije Universirteit Brussel)
+--- run_braker.py.orig	2023-04-03 12:07:35.178272000 +0200
++++ run_braker.py	2023-04-03 12:24:48.713953000 +0200
+@@ -124,8 +124,19 @@
+         logger_time.debug('START: BRAKER')
+
+         if not os.path.exists(gff3_braker):
+-            augustus_config_path = os.path.join(
+-                os.path.dirname(D_CONF['AUGUSTUS_PATH']), '../config')
++            augustus_bin_path = os.environ[
++                'AUGUSTUS_BIN_PATH'
++            ] or os.path.dirname(D_CONF['AUGUSTUS_PATH'])
++            augustus_config_path = os.environ[
++                'AUGUSTUS_CONFIG_PATH'
++            ] or os.path.join(
++                os.path.dirname(D_CONF['AUGUSTUS_PATH']), '../config'
++            )
++            augustus_scripts_path = os.environ[
++                'AUGUSTUS_SCRIPTS_PATH'
++            ] or os.path.join(
++                os.path.dirname(D_CONF['AUGUSTUS_PATH']), '../scripts'
++            )
+             config_species = os.path.join(
+                 augustus_config_path, 'species', prefix)
+             species = prefix
+@@ -140,7 +151,6 @@
+             bamtools_path = os.path.dirname(D_CONF['BAMTOOLS_PATH'])
+             genemark_path = os.path.dirname(D_CONF['GENEMARK_PATH'])
+             samtools_path = os.path.dirname(D_CONF['SAMTOOLS_PATH'])
+-            augustus_scripts_path = os.path.dirname(D_CONF['AUGUSTUS_PATH'])
+             working_dir = os.path.join(output_dir, prefix)
+             if not os.path.exists(working_dir):
+                 os.mkdir(working_dir)
+@@ -153,7 +163,7 @@
+                     braker_bin, fungus_flag, num_cores, adjusted_assembly,
+                     bam_file, species, augustus_config_path, bamtools_path,
+                     genemark_path, samtools_path, working_dir,
+-                    translation_table, augustus_scripts_path, log_braker))
++                    translation_table, augustus_bin_path, log_braker))
+             logger_txt.debug('[Run] %s', command1)
+             os.system(command1)
+
+@@ -165,7 +175,7 @@
+
+             augustus_dir = os.path.dirname(D_CONF['AUGUSTUS_PATH'])
+             get_anno_script = os.path.join(
+-                augustus_dir, 'getAnnoFastaFromJoingenes.py')
++                augustus_scripts_path, 'getAnnoFastaFromJoingenes.py')
+             if not os.path.exists(get_anno_script):
+                 get_anno_script = os.path.join(
+                     augustus_dir,

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1_fix-maker-exes.patch

@@ -0,0 +1,20 @@
+The paths to MAKER executables set by MAKER are already correct, avoid
+tampering with them.
+author: Alex Domingo (Vrije Universiteit Brussel)
+--- run_maker.py.orig	2023-03-21 09:19:05.057679000 +0100
++++ run_maker.py	2023-03-21 09:19:49.143408491 +0100
+@@ -463,14 +463,6 @@
+     replace('maker_opts.ctl', 'protein_pass=0', 'protein_pass=1')
+     replace('maker_opts.ctl', 'rm_pass=0', 'rm_pass=1')
+
+-    # Program paths
+-    for program in [
+-            'makeblastdb', 'blastn', 'blastx', 'tblastx', 'RepeatMasker',
+-            'exonerate', 'snap', 'augustus', 'tRNAscan-SE', 'snoscan']:
+-        replace('maker_exe.ctl', '{}='.format(program), '{}={}'.format(
+-            program, os.path.join(os.path.dirname(maker_bin), program)
+-        ))
+-
+     # Last run, keep_preds=1
+     if version == '4':
+         replace('maker_opts.ctl', 'keep_preds=0', 'keep_preds=1')

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1_fix-snap-detection.patch

@@ -0,0 +1,61 @@
+Fix detection of fathom, forge and hmm-assembler.pl, which are not part of MAKER but SNAP
+see https://github.com/CompSynBioLab-KoreaUniv/FunGAP/pull/97
+author: Alex Domingo (Vrije Universiteit Brussel)
+--- set_dependencies.py.orig	2023-03-16 15:27:00.506475000 +0100
++++ set_dependencies.py	2023-03-16 15:31:31.102674857 +0100
+@@ -44,6 +44,10 @@
+         help='Maker bin path'
+     )
+     parser.add_argument(
++        '-s', '--snap_path', nargs=1, required=True,
++        help='SNAP-HMM bin path'
++    )
++    parser.add_argument(
+         '-r', '--with_repeat_modeler', nargs='?', default='',
+         help='User-defined RepeatModeler bin path'
+     )
+@@ -76,6 +80,7 @@
+     pfam_db_path = os.path.abspath(args.pfam_db_path[0])
+     i_genemark_path = os.path.abspath(args.genemark_path[0])
+     i_maker_path = os.path.abspath(args.maker_path[0])
++    i_snap_path = os.path.abspath(args.snap_path[0])
+     if args.with_repeat_modeler:
+         with_repeat_modeler = os.path.abspath(args.with_repeat_modeler)
+     else:
+@@ -104,8 +109,9 @@
+         pfam_scan_path, blastp_path, blastn_path, blastx_path,
+         makeblastdb_path, samtools_path, bamtools_path, augustus_path
+     ) = get_path(
+-        i_genemark_path, i_maker_path, with_repeat_modeler, with_augustus,
+-        with_hisat2, with_trinity, with_braker, with_busco, with_pfam_scan
++        i_genemark_path, i_maker_path, i_snap_path, with_repeat_modeler,
++        with_augustus, with_hisat2, with_trinity, with_braker, with_busco,
++        with_pfam_scan
+     )
+     check_working(
+         genemark_path, gmhmme3_path, probuild_path, build_database_path,
+@@ -159,8 +165,9 @@
+
+
+ def get_path(
+-        i_genemark_path, i_maker_path, with_repeat_modeler, with_augustus,
+-        with_hisat2, with_trinity, with_braker, with_busco, with_pfam_scan):
++        i_genemark_path, i_maker_path, i_snap_path, with_repeat_modeler,
++        with_augustus, with_hisat2, with_trinity, with_braker, with_busco,
++        with_pfam_scan):
+     '''Get path'''
+     print('\n** Checking the installed locations of dependencies **\n')
+
+@@ -198,9 +205,9 @@
+     gff3_merge_path = check_binary('Maker', i_maker_path, 'gff3_merge')
+     fasta_merge_path = check_binary('Maker', i_maker_path, 'fasta_merge')
+     maker2zff_path = check_binary('Maker', i_maker_path, 'maker2zff')
+-    fathom_path = check_binary('Snap', i_maker_path, 'fathom')
+-    forge_path = check_binary('Snap', i_maker_path, 'forge')
+-    hmm_assembler_path = check_binary('Snap', i_maker_path, 'hmm-assembler.pl')
++    fathom_path = check_binary('Snap', i_snap_path, 'fathom')
++    forge_path = check_binary('Snap', i_snap_path, 'forge')
++    hmm_assembler_path = check_binary('Snap', i_snap_path, 'hmm-assembler.pl')
+     build_database_path = check_binary(
+         'RepeatModeler (BuildDatabase)', with_repeat_modeler, 'BuildDatabase',
+     )

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1_relax-dependency-checks.patch

@@ -0,0 +1,29 @@
+Disable too strict check on version of AUGUSTUS and the GeneMark key in the
+home directory of the user
+author: Alex Domingo (Vrije Universiteit Brusel)
+--- set_dependencies.py.orig	2023-03-16 15:50:03.802522000 +0100
++++ set_dependencies.py	2023-03-16 15:50:28.562603082 +0100
+@@ -284,15 +284,15 @@
+     check_working_internal(samtools_path, [samtools_path, '--help'])
+     check_working_internal(bamtools_path, [bamtools_path, '--help'])
+     check_working_internal(augustus_path, [augustus_path, '--help'])
+-    check_augustus_version(augustus_path)
++    # check_augustus_version(augustus_path)
+
+-    # For GeneMark, check the .gm_key
+-    home_path = os.path.expanduser('~')
+-    if not os.path.exists(os.path.join(home_path, '.gm_key')):
+-        sys.exit(
+-            '\n[ERROR] You do not have .gm_key in your home directory.\n'
+-            'Check https://wiki.gacrc.uga.edu/wiki/GeneMark'
+-        )
++    # # For GeneMark, check the .gm_key
++    # home_path = os.path.expanduser('~')
++    # if not os.path.exists(os.path.join(home_path, '.gm_key')):
++    #     sys.exit(
++    #         '\n[ERROR] You do not have .gm_key in your home directory.\n'
++    #         'Check https://wiki.gacrc.uga.edu/wiki/GeneMark'
++    #     )
+
+ def check_augustus_version(augustus_path):
+     '''Check Augustus version 3.3.3'''

easybuild/easyconfigs/f/FunGAP/FunGAP-1.1.1_replace-deprecated-pa-repeatmodeler.patch

@@ -0,0 +1,22 @@
+Replace deprecated -pa option in newer versions of RepeatModeler with -threads
+author: Alex Domingo (Vrije Universiteit Brussel)
+--- run_repeat_modeler.py.orig	2023-03-17 09:23:21.857374000 +0100
++++ run_repeat_modeler.py	2023-03-17 09:24:17.655942000 +0100
+@@ -73,7 +73,7 @@
+
+     # BuildDatabase -name Choanephora_cucurbitarum
+     # ../Choanephora_cucurbitarum_assembly.fna
+-    # RepeatModeler -database Choanephora_cucurbitarum -pa 25
++    # RepeatModeler -database Choanephora_cucurbitarum -threads 25
+
+     # Get repeat model
+     repeat_lib = os.path.join(output_dir, '*', 'consensi.fa.classified')
+@@ -89,7 +89,7 @@
+         os.system(command1)
+
+         log_file2 = os.path.join(log_dir, 'repeat_modeler.log')
+-        command2 = '{} -database {} -pa {} > {} 2>&1'.format(
++        command2 = '{} -database {} -threads {} > {} 2>&1'.format(
+             repeatmodeler_bin, genome_assembly, num_cores, log_file2
+         )
+         logger_txt.debug('[Run] %s', command2)