stage for v1.6.0 release

Author: Jon Palmer

bin/funannotate-test.py

@@ -17,7 +17,7 @@ def __init__(self,prog):
     epilog="""Written by Jon Palmer (2016-2018) [email protected]""",
     formatter_class = MyFormatter)
 parser.add_argument('-t','--tests', required=True, nargs='+',
-                    choices=['all', 'clean', 'mask', 'predict', 'annotate', 'busco', 'rna-seq', 'compare'],
+                    choices=['all', 'clean', 'mask', 'predict', 'annotate', 'rna-seq', 'compare'],
                     help='select which tests to run')
 parser.add_argument('--cpus', default=2, type=int, help='Number of CPUs to use')                    
 args=parser.parse_args()
@@ -165,7 +165,7 @@ def runPredictTest():
     print("#########################################################")
     #check results
     try:
-        assert 1500 <= countGFFgenes(os.path.join(tmpdir, 'annotate', 'predict_results', 'Awesome_testicus.gff3')) <= 1700
+        assert 1500 <= countGFFgenes(os.path.join(tmpdir, 'annotate', 'predict_results', 'Awesome_testicus.gff3')) <= 1800
         print('SUCCESS: `funannotate predict` test complete.')
         shutil.rmtree(tmpdir)
     except AssertionError:
@@ -182,7 +182,7 @@ def runBuscoTest():
         lib.log.error("$AUGUSTUS_CONFIG_PATH environmental variable not found, set to continue.")
         return
     if os.path.isdir(os.path.join(AUGUSTUS, 'species', 'awesome_busco')):
-    	shutil.rmtree(os.path.join(AUGUSTUS, 'species', 'awesome_busco'))
+        shutil.rmtree(os.path.join(AUGUSTUS, 'species', 'awesome_busco'))
     tmpdir = 'test-busco_'+pid
     os.makedirs(tmpdir)
     inputFasta = 'test.softmasked.fa'
@@ -201,7 +201,7 @@ def runBuscoTest():
     print("#########################################################")
     #check results
     try:
-        assert 1500 <= countGFFgenes(os.path.join(tmpdir, 'annotate', 'predict_results', 'Awesome_busco.gff3')) <= 1700
+        assert 1500 <= countGFFgenes(os.path.join(tmpdir, 'annotate', 'predict_results', 'Awesome_busco.gff3')) <= 1800
         print('SUCCESS: `funannotate predict` BUSCO-mediated training test complete.')
         shutil.rmtree(tmpdir)
     except AssertionError:
@@ -259,7 +259,7 @@ def runCompareTest():
     shutil.copyfile(input3, os.path.join(tmpdir, input3))
     #run predict
     runCMD(['funannotate', 'compare', 
-    		'-i', input1, input2, input3,
+            '-i', input1, input2, input3,
             '-o', 'compare', '--cpus', str(args.cpus), 
             '--run_dnds', 'estimate', '--outgroup', 'botrytis_cinerea.dikarya'], tmpdir)
     print("#########################################################")
@@ -273,7 +273,7 @@ def runCompareTest():
     except AssertionError:
         print('ERROR: `funannotate compare` test failed - check logfiles')
     print("#########################################################\n")
-	
+    
 def runRNAseqTest():
     print("#########################################################")
     print('Running funannotate RNA-seq training/prediction unit testing')
@@ -284,7 +284,7 @@ def runRNAseqTest():
         lib.log.error("$AUGUSTUS_CONFIG_PATH environmental variable not found, set to continue.")
         return
     if os.path.isdir(os.path.join(AUGUSTUS, 'species', 'awesome_rna')):
-    	shutil.rmtree(os.path.join(AUGUSTUS, 'species', 'awesome_rna'))
+        shutil.rmtree(os.path.join(AUGUSTUS, 'species', 'awesome_rna'))
     tmpdir = 'test-rna_seq_'+pid
     os.makedirs(tmpdir)
     inputFasta = 'test.softmasked.fa'
@@ -296,7 +296,7 @@ def runRNAseqTest():
             download(download_links.get('rna-seq'), 'test-rna_seq.tar.gz')
         subprocess.call(['tar', '-zxf', 'test-rna_seq.tar.gz'])
     for f in [inputFasta, protEvidence, illumina, nanopore]:
-    	shutil.copyfile(f, os.path.join(tmpdir, f))
+        shutil.copyfile(f, os.path.join(tmpdir, f))
     #run train
     runCMD(['funannotate', 'train', '-i', inputFasta, 
             '--single', illumina, '--nanopore_mrna', nanopore,
@@ -313,7 +313,7 @@ def runRNAseqTest():
     print("#########################################################")
     print('Now running `funannotate update` to run PASA-mediated UTR addition and multiple transcripts')
     runCMD(['funannotate', 'update', '-i', 'rna-seq', 
-			'--cpus', str(args.cpus)], tmpdir)    
+            '--cpus', str(args.cpus)], tmpdir)    
     print("#########################################################")
     #check results
     try:
@@ -332,11 +332,11 @@ def runRNAseqTest():
     runMaskTest()
 if 'predict' in args.tests or 'all' in args.tests:
     runPredictTest()
-if 'busco' in args.tests or 'all' in args.tests:
-    runBuscoTest()
+#if 'busco' in args.tests or 'all' in args.tests:
+#    runBuscoTest()
 if 'rna-seq' in args.tests or 'all' in args.tests:
-	runRNAseqTest()
+    runRNAseqTest()
 if 'annotate' in args.tests or 'all' in args.tests:
-	runAnnotateTest()
+    runAnnotateTest()
 if 'compare' in args.tests or 'all' in args.tests:
-	runCompareTest()
+    runCompareTest()

dockerbuild/Dockerfile-base

@@ -31,7 +31,7 @@ RUN wget --quiet https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86
     sudo /bin/bash ~/miniconda.sh -b -p /home/linuxbrew/conda &&  rm ~/miniconda.sh && sudo chown -R linuxbrew: /home/linuxbrew/conda && \
     conda update -y conda && conda config --add channels r && conda config --add channels defaults && \
     conda config --add channels etetoolkit && conda config --add channels conda-forge && conda config --add channels bioconda && \
-    conda install -y numpy pandas scipy matplotlib seaborn natsort scikit-learn psutil biopython requests \
+    conda install -y numpy pandas scipy matplotlib seaborn natsort scikit-learn psutil biopython requests snap glimmerhmm=3.0.4=pl526h0a44026_3 \
     goatools fisher bedtools blat hmmer exonerate diamond tbl2asn hisat2 ucsc-pslcdnafilter stringtie perl-dbd-sqlite \
     samtools raxml trimal mafft phyml kallisto bowtie2 infernal perl-threaded perl-db-file perl-bioperl perl-dbd-mysql \
     perl-app-cpanminus mummer ete3 ete_toolchain minimap2 salmon=0.9.1 jellyfish htslib nano perl-text-soundex perl-scalar-util-numeric && \
@@ -100,5 +100,5 @@ RUN wget --quiet https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86
 
 COPY pasa_conf.txt /home/linuxbrew/PASApipeline/pasa_conf/conf.txt
 
-RUN wget https://github.com/nextgenusfs/funannotate/archive/1.5.3.tar.gz && \
-    tar -zxvf 1.5.3.tar.gz && rm 1.5.3.tar.gz && mv funannotate-1.5.3 funannotate
+RUN wget https://github.com/nextgenusfs/funannotate/archive/1.6.0.tar.gz && \
+    tar -zxvf 1.6.0.tar.gz && rm 1.6.0.tar.gz && mv funannotate-1.6.0 funannotate

funannotate.py

@@ -421,7 +421,7 @@ def fmtcols(mylist, cols):
              you cannot download from the machine funannotate is installed at - then download
              the 7 tar.gz files from https://osf.io/bj7v4/files/ and run script from directory 
     
-Arguments:   -t, --tests    Test sets to run. [all,clean,mask,predict,busco,rna-seq,annotate,compare]
+Arguments:   -t, --tests    Test sets to run. [all,clean,mask,predict,rna-seq,annotate,compare]
              --cpus         Number of cpus to use. Default: 2
              
 Written by Jon Palmer (2016-2018) [email protected]

util/check_modules.py

@@ -91,6 +91,8 @@ def check_version2(name):
         elif name == 'mafft':
             vers = subprocess.Popen([name, '--version'], stdout=subprocess.PIPE, stderr=subprocess.PIPE).communicate()[1]
             vers = vers.strip()
+        elif name == 'emapper.py':
+            vers = subprocess.Popen([name, '--version'], stdout=subprocess.PIPE).communicate()[0].split('\n')[1].replace('emapper-', '')
         else:
             vers = subprocess.Popen([name, '--version'], stdout=subprocess.PIPE).communicate()[0].split('\n')[0]
         if 'exonerate' in vers:
@@ -158,11 +160,19 @@ def check_version5(name):
             vers = subprocess.Popen([name], stdout=subprocess.PIPE, stderr=subprocess.PIPE).communicate()
             vers = 'no way to determine'
         elif name == 'CodingQuarry':
-        	vers = subprocess.Popen([name], stdout=subprocess.PIPE).communicate()
-        	v = vers[0].split('\n')
-        	for i in v:
-        		if 'CodingQuarry v.' in i:
-        			vers = i.split('v. ')[-1]
+            vers = subprocess.Popen([name], stdout=subprocess.PIPE).communicate()
+            v = vers[0].split('\n')
+            for i in v:
+                if 'CodingQuarry v.' in i:
+                    vers = i.split('v. ')[-1]
+        elif name == 'snap':
+            vtmp = subprocess.Popen(['snap'], stderr=subprocess.STDOUT, stdout=subprocess.PIPE).communicate()[0].rstrip().split('\n')
+            for i in vtmp:
+                if i.startswith('SNAP'):
+                    vers = i.split('(version ')[-1].rstrip(')')
+        elif name == 'glimmerhmm':
+            vtmp = subprocess.Popen(['snap'], stderr=subprocess.STDOUT, stdout=subprocess.PIPE).communicate()[0].rstrip().split('\n')
+            vers = '3.0.4'
     except OSError as e:
         if e.errno == os.errno.ENOENT:
             return False
@@ -207,12 +217,12 @@ def check_version6(name):
 
 programs1 = ['tblastn', 'makeblastdb', 'rmblastn', 'java'] #-version
 programs2 = ['exonerate', 'bedtools', 'bamtools', 'augustus', 
-			 'samtools', 'gmap', 'hisat2', 'Trinity', 'nucmer', 
-			 'tbl2asn', 'emapper.py', 'minimap2', 'mafft', 
-			 'trimal', 'stringtie'] #--version
+             'samtools', 'gmap', 'hisat2', 'Trinity', 'nucmer', 
+             'tbl2asn', 'emapper.py', 'minimap2', 'mafft', 
+             'trimal', 'stringtie', 'salmon'] #--version
 programs3 = ['RepeatModeler', 'RepeatMasker'] #-v
 programs4 = ['diamond', 'ete3', 'kallisto'] #version
-programs5 = ['gmes_petap.pl', 'blat', 'pslCDnaFilter', 'fasta', 'CodingQuarry'] #no version option at all, a$$holes
+programs5 = ['gmes_petap.pl', 'blat', 'pslCDnaFilter', 'fasta', 'CodingQuarry', 'snap', 'glimmerhmm'] #no version option at all, a$$holes
 programs6 = ['hmmsearch', 'hmmscan', 'tRNAscan-SE'] #-h
 programs7 = ['signalp'] # -V