updates to v0.1.5

Author: Jon Palmer

bin/funannotate-compare.py

@@ -87,8 +87,8 @@ def __init__(self,prog):
 
 
 #check dependencies and set path to proteinortho
-PROTORTHO = os.path.join(parentdir, 'util', 'proteinortho_v5.11', 'proteinortho5.pl')
-programs = ['find_enrichment.py', 'mafft', 'raxmlHPC-PTHREADS', 'trimal']
+#PROTORTHO = os.path.join(parentdir, 'util', 'proteinortho_v5.11', 'proteinortho5.pl')
+programs = ['find_enrichment.py', 'mafft', 'raxmlHPC-PTHREADS', 'trimal', 'proteinortho5.pl']
 lib.CheckDependencies(programs)
 
 #copy over html files
@@ -461,7 +461,7 @@ def __init__(self,prog):
         if file.endswith('.faa'):
             filelist.append(file)
     fileinput = ' '.join(filelist)
-    cmd = [PROTORTHO, '-project=funannotate', '-synteny', '-cpus='+str(args.cpus), '-singles', '-selfblast']
+    cmd = ['proteinortho5.pl', '-project=funannotate', '-synteny', '-cpus='+str(args.cpus), '-singles', '-selfblast']
     cmd2 = cmd + filelist
     if not os.path.isfile(os.path.join(args.out, 'protortho', 'funannotate.poff')):
         with open(log, 'w') as logfile:
@@ -706,6 +706,8 @@ def addlink(x):
     output.write('<p>Report generated on: '+ d[1]+'/'+d[2]+'/'+d[0]+ ' at '+str(hour)+':'+d[4]+ ' '+m+'</p>')
     output.write(lib.FOOTER)
 
+lib.log.info("Compressing results to output file: %s.tar.gz" % args.out)
+lib.make_tarfile(args.out+'.tar.gz', args.out)
 lib.log.info("Finished!")
 os._exit(1)
 

bin/funannotate-predict.py

@@ -52,6 +52,9 @@
 del arguments[index]
 del arguments[index]
 cmd1 = base_cmd1 + arguments
+#remove intron from partitions command
+del cmd1[-1]
+del cmd1[-1]
 
 
 def grouper(n, iterable, fillvalue=None):
@@ -74,6 +77,8 @@ def safe_run(*args, **kwargs):
 #split partitions
 lib.log.info("Setting up EVM partitions")
 subprocess.call(cmd1, cwd = tmpdir, stdout = FNULL, stderr = FNULL)
+#check output
+lib.checkinputs(os.path.join(tmpdir, 'partitions_list.out'))
 
 #generate commands
 lib.log.info("Generating EVM command list")

bin/funannotate-runEVM.py

@@ -31,7 +31,7 @@ def fmtcols(mylist, cols):
              for i in range(0,num_lines))
     return "\n".join(lines)
 
-version = '0.1.4'
+version = '0.1.5'
 
 default_help = """
 Usage:       funannotate <command> <arguments>
@@ -121,6 +121,7 @@ def fmtcols(mylist, cols):
              
 Optional:  --isolate              Strain isolate, e.g. Af293            
            --name                 Locus tag name (assigned by NCBI?). Default: FUN_
+           --maker_gff            MAKER2 GFF file. Parse results directly to EVM.
            --pasa_gff             PASA generated gene models
            --rna_bam              RNA-seq mapped to genome to train Augustus/GeneMark-ET       
            --augustus_species     Augustus species config. Default: uses species name

funannotate.py

@@ -11,6 +11,13 @@ command -v makeblastdb >/dev/null 2>&1 || { echo "Funannotate setup requires BLA
 
 dir=$PWD
 outputdir='/usr/local/share/funannotate'
+echo -n "DB directory set to ($outputdir), continue [y/n]: "
+read question1
+if [ $question1 == 'n' ]; then
+    echo -n "Enter path to DB directory: "
+    read dbname
+    outputdir=$dbname
+fi
 
 if [ -z "$1" ]; then
 	db='pass'
@@ -194,20 +201,12 @@ if [ "$dep" = 'pass' ]; then
     echo "-----------------------------------------------"
 
     #setup some programs and look for dependencies
-    #do some OS check and rename proteinortho clustering binary accordingly (this was difficult to compile on mac, hopefully this saves everybody the trouble)
 
     #make sure in funannotate directory
     cd $dir
-    if ! [ -f util/proteinortho_v5.11/proteinortho5_clustering ]; then
-        if [[ $OSTYPE == darwin* ]]; then
-            cp util/proteinortho_v5.11/proteinortho5_clustering_osx util/proteinortho_v5.11/proteinortho5_clustering
-        else
-            sudo cp util/proteinortho_v5.11/proteinortho5_clustering_linux util/proteinortho_v5.11/proteinortho5_clustering
-        fi
-    fi
 
     check='pass'
-    for i in {blastp,hmmsearch,hmmscan,augustus,'gmes_petap.pl',mummer,nucmer,show-coords,exonerate,gmap,blat,RepeatModeler,RepeatMasker,pslCDnaFilter,bedtools,bamtools,'gag.py',tbl2asn,'braker.pl',funannotate,mafft,trimal,raxmlHPC-PTHREADS,tRNAscan-SE,'rmOutToGFF3.pl'}; do
+    for i in {blastp,hmmsearch,hmmscan,augustus,'gmes_petap.pl',mummer,nucmer,show-coords,exonerate,gmap,blat,RepeatModeler,RepeatMasker,pslCDnaFilter,bedtools,bamtools,'gag.py',tbl2asn,'braker.pl',funannotate,mafft,trimal,raxmlHPC-PTHREADS,tRNAscan-SE,'rmOutToGFF3.pl','proteinortho5.pl'}; do
         var=$(command -v $i)
         if [ "$var" ]; then
             echo "$i installed.........$var"
@@ -259,3 +258,18 @@ if [ "$dep" = 'pass' ]; then
         echo -e "Script complete, funannotate is ready to roll!\n"
     fi
 fi
+
+#vertebrate EggNog
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/veNOG/veNOG.hmm.tar.gz
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/veNOG/veNOG.annotations.tsv.gz
+#http://busco.ezlab.org/files/vertebrata_buscos.tar.gz
+
+#arthropods
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/artNOG/artNOG.hmm.tar.gz
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/artNOG/artNOG.annotations.tsv.gz
+#http://busco.ezlab.org/files/arthropoda_buscos.tar.gz
+
+#metazoans
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/meNOG/meNOG.hmm.tar.gz
+#http://eggnogdb.embl.de/download/eggnog_4.5/data/meNOG/meNOG.annotations.tsv.gz
+#http://busco.ezlab.org/files/metazoa_buscos.tar.gz

lib/library.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env perl -w
+use strict;
+use warnings;
+
+#written by Jason Stajich
+#https://github.com/hyphaltip/genome-scripts/blob/master/gene_prediction/maker2evm.pl
+
+open(my $tr => ">transcript_alignments.gff3")|| die $!;
+open(my $gene => ">gene_predictions.gff3")|| die $!;
+open(my $pep => ">protein_alignments.gff3")|| die $!;
+
+while(<>) {
+    if (/^\#/) {
+	next;
+    } elsif (/^>/ ) {
+	last;
+    }
+    chomp;
+    my @row = split(/\t/,$_);
+    if( $row[1] =~ /genemark|snap_masked|augustus_masked/ ) {
+	$row[1] =~ s/_masked//;
+	if( $row[2] eq 'match' ) {	    
+	    $row[2] = 'gene';
+	    print $gene join("\t",@row),"\n";
+	    $row[2] = 'mRNA';
+	} elsif( $row[2] eq 'match_part' ) {
+	    $row[2] = 'CDS';
+	} else {
+	    warn("unknown field type for $row[2]\n");
+	}
+	print $gene join("\t", @row), "\n";
+    } elsif( $row[1] eq 'est2genome' ) {
+	next if $row[2] eq 'expressed_sequence_match';
+	if( $row[2] eq 'match_part' ) {
+	    $row[2] = 'EST_match';
+	}
+	print $tr join("\t", @row), "\n";
+    } elsif( $row[1] eq 'protein2genome' ) {
+	next if $row[2] eq 'protein_match';
+	if( $row[2] eq 'match_part' ) {
+	    $row[2] = 'nucleotide_to_protein_match';
+	}
+	print $pep join("\t", @row), "\n";
+    } elsif( $row[1] =~ /maker|repeatmasker|blast[xn]|repeatrunner|\./) {
+	next; # skipping these
+    } else {
+	warn("unknown type for $row[1] $row[2]\n");
+	next;
+    }
+#    print join("\t", @row),"\n";
+}