Make sspace and gapfiller optional

README.md

@@ -13,13 +13,17 @@ Installation
 ARIBA has the following dependencies, which need to be installed:
   * [cd-hit] [cdhit] version >= 4.6
   * [samtools and bcftools] [samtools]  version >= 1.2
-  * [SSPACE-basic scaffolder] [sspace]
-  * [GapFiller] [gapfiller]
   * [MUMmer] [mummer] version >= 3.23
   * [SMALT] [smalt] version >= 0.7.4
   * Either [SPAdes] [spades] version >= 3.5.0 or [Velvet] [velvet] version >= 1.2.07
     (SPAdes is recommended)
 
+ARIBA has the following optional dependencies. If they are installed,
+they will be used. Otherwise scaffolding and gap filling will be
+skipped.
+  * [SSPACE-basic scaffolder] [sspace]
+  * [GapFiller] [gapfiller]
+
 Once the dependencies are installed, install ARIBA using pip:
 
     pip3 install pyfastaq

ariba/cluster.py

@@ -80,19 +80,18 @@ def __init__(self,
 
         self.bcftools_exe = bcftools_exe
 
-        self.gapfiller_exe = shutil.which(gapfiller_exe)
-        if self.gapfiller_exe is None:
-            raise Error('Error! ' + gapfiller_exe + ' not found in path')
-        self.gapfiller_exe = os.path.realpath(self.gapfiller_exe) # otherwise gapfiller dies loading packages
+        self.sspace_exe = shutil.which(sspace_exe)
+        if self.sspace_exe is None:
+            self.gapfiller_exe = None
+        else:
+            self.sspace_exe = os.path.realpath(self.sspace_exe) # otherwise sspace dies loading packages
+            self.gapfiller_exe = shutil.which(gapfiller_exe)
+            if self.gapfiller_exe is not None:
+                self.gapfiller_exe = os.path.realpath(self.gapfiller_exe) # otherwise gapfiller dies loading packages
 
         self.samtools_exe = samtools_exe
         self.smalt_exe = smalt_exe
 
-        self.sspace_exe = shutil.which(sspace_exe)
-        if self.sspace_exe is None:
-            raise Error('Error! ' + sspace_exe + ' not found in path')
-        self.sspace_exe = os.path.realpath(self.sspace_exe) # otherwise sspace dies loading packages
-
         if self.assembler == 'velvet':
             self.velveth = velvet_exe + 'h'
             self.velvetg = velvet_exe + 'g'
@@ -305,6 +304,13 @@ def _scaffold_with_sspace(self):
             raise Error('Error mkdir '+  self.scaffold_dir)
 
         cwd = os.getcwd()
+
+        if self.sspace_exe is None:
+            os.chdir(self.assembly_dir)
+            os.symlink(os.path.basename(self.assembly_contigs), os.path.basename(self.scaffolder_scaffolds))
+            os.chdir(cwd)
+            return
+
         os.chdir(self.scaffold_dir)
         lib_file = 'lib'
         with open(lib_file, 'w') as f:
@@ -339,7 +345,7 @@ def _gap_fill_with_gapfiller(self):
 
         cwd = os.getcwd()
 
-        if not self._has_gaps_to_fill(self.scaffolder_scaffolds):
+        if self.gapfiller_exe is None or not self._has_gaps_to_fill(self.scaffolder_scaffolds):
             self._rename_scaffolds(self.scaffolder_scaffolds, self.gapfilled_scaffolds)
             return
 

ariba/clusters.py

@@ -88,19 +88,21 @@ def __init__(self,
 
         self.bcftools_exe = bcftools_exe
 
-        self.gapfiller_exe = shutil.which(gapfiller_exe)
-        if self.gapfiller_exe is None:
-            raise Error('Error! ' + gapfiller_exe + ' not found in path')
-        self.gapfiller_exe = os.path.realpath(self.gapfiller_exe) # otherwise gapfiller dies loading packages
+        self.sspace_exe = shutil.which(sspace_exe)
+        if self.sspace_exe is None:
+            print('WARNING: SSPACE not found. Scaffolding and gap filling will be skipped!', file=sys.stderr)
+            self.gapfiller_exe = None
+        else:
+            self.sspace_exe = os.path.realpath(self.sspace_exe) # otherwise sspace dies loading packages
+            self.gapfiller_exe = shutil.which(gapfiller_exe)
+            if self.gapfiller_exe is None:
+                print('WARNING: GapFiller not found. No gap filling will be run after scaffolding!', file=sys.stderr)
+            else:
+                self.gapfiller_exe = os.path.realpath(self.gapfiller_exe) # otherwise gapfiller dies loading packages
 
         self.samtools_exe = samtools_exe
         self.spades_exe = spades_exe
 
-        self.sspace_exe = shutil.which(sspace_exe)
-        if self.sspace_exe is None:
-            raise Error('Error! ' + sspace_exe + ' not found in path')
-        self.sspace_exe = os.path.realpath(self.sspace_exe) # otherwise sspace dies loading packages
-
         self.velvet = velvet_exe
 
         self.cdhit_seq_identity_threshold = cdhit_seq_identity_threshold

ariba/external_progs.py

@@ -74,13 +74,16 @@ def set_path(prog, opts):
     exec('opts.' + prog + ' = "' + prog_to_default[prog] + '"')
 
 
-def get_version(prog, path=None):
+def get_version(prog, path=None, raise_error=True):
     assert prog in prog_to_version_cmd
     if path is None:
         path = prog
 
     if not is_in_path(path):
-        raise Error('Error getting version of ' + path + ' - not found in path.')
+        if raise_error:
+            raise Error('Error getting version of ' + path + ' - not found in path.')
+        else:
+            return 'Not_in_path', 'Not_in_path'
 
     path = shutil.which(path)
 
@@ -100,7 +103,10 @@ def get_version(prog, path=None):
     return 'UNKNOWN ...\n I tried running this to get the version: "' + cmd + '"\n and the output didn\'t match this regular expression: "' + regex.pattern + '"', path
 
 
-def check_versions(opts, verbose=False):
+def check_versions(opts, verbose=False, not_required=None):
+    if not_required is None:
+        not_required = set()
+
     if verbose:
         print('{:_^79}'.format(' Checking dependencies and their versions '))
         print('tool', 'version', 'path', sep='\t')
@@ -124,22 +130,33 @@ def check_versions(opts, verbose=False):
         raise Error('Assembler ' + opts.assembler + ' not recognised. Cannot continue')
 
     errors = []
+    failed_to_find = set()
 
     for prog in to_check:
         set_path(prog, opts)
-        version, path = get_version(prog, path=eval('opts.' + prog))
+        version, path = get_version(prog, path=eval('opts.' + prog), raise_error=prog not in not_required)
         if verbose:
             print(prog, version, path, sep='\t')
+        if path == 'Not_in_path':
+            print('\nWARNING:', prog, 'not found in path, so will be skipped during assembly\n', file=sys.stderr)
 
         if prog in min_versions and LooseVersion(version) < LooseVersion(min_versions[prog]):
             errors.append(' '.join(['Found version', version, 'of', prog, 'which is too low! Please update to at least', min_versions[prog] + '\n   Found it here:', path]))
+            failed_to_find.add(prog)
 
     if len(errors):
         for e in errors:
             print('\n*** Error! Bad dependency! ***', file=sys.stderr)
             print(e, file=sys.stderr)
             print()
-        raise Error('Cannot continue. Some dependencies need updating')
+        if len(failed_to_find.difference(not_required)) > 0:
+            raise Error('Cannot continue. Some dependencies need updating')
+        else:
+            assert failed_to_find.issubset(not_required)
+            if 'sspace' in failed_to_find:
+                print('WARNING: SSPACE not found. Will not run scaffolding or gap filling', file=sys.stderr)
+            elif 'gapfiller' in failed_to_find:
+                print('WARNING: GapFiller not found. Will not run gap filling after scaffolding', file=sys.stderr)
 
     if verbose:
-        print('\n... dependencies look OK.\n')
+        print('\nDependencies look OK (but check in case there are warnings about SSPACE or GapFiller)\n')

ariba/tasks/run.py

@@ -57,7 +57,7 @@ def run():
     options = parser.parse_args()
     if options.assembler == 'velvet':
         options.velvet = 'velvet'
-    ariba.external_progs.check_versions(options, verbose=options.verbose)
+    ariba.external_progs.check_versions(options, verbose=options.verbose, not_required=set(['sspace', 'gapfiller']))
     pyfastaq.sequences.genetic_code = options.genetic_code
 
     c = ariba.clusters.Clusters(