Merge pull request #53 from martinghunt/expose_spades_careful_only_as…

…semble

Expose spades careful, only-assembler

circlator/assemble.py

@@ -13,6 +13,8 @@ def __init__(self,
       outdir,
       threads=1,
       spades_kmers=None,
+      careful=True,
+      only_assembler=True,
       verbose=False,
       spades_use_first_success=False,
     ):
@@ -23,6 +25,8 @@ def __init__(self,
 
         self.verbose = verbose
         self.threads = threads
+        self.careful = careful
+        self.only_assembler = only_assembler
         self.spades = external_progs.make_and_check_prog('spades', verbose=self.verbose)
         self.spades_kmers = self._build_spades_kmers(spades_kmers)
         self.spades_use_first_success = spades_use_first_success
@@ -45,17 +49,26 @@ def _build_spades_kmers(self, kmers):
             raise Error('Error getting list of kmers from:' + str(kmers))
 
 
-    def run_spades_once(self, kmer, outdir):
-        cmd = ' '.join([
+    def _make_spades_command(self, kmer, outdir):
+        cmd = [
             self.spades.exe(),
             '-s', self.reads,
-            '-k', str(kmer),
-            '--careful',
-            '--only-assembler',
-            '-t', str(self.threads),
             '-o', outdir,
-        ])
+            '-t', str(self.threads),
+            '-k', str(kmer),
+        ]
 
+        if self.careful:
+            cmd.append('--careful')
+
+        if self.only_assembler:
+            cmd.append('--only-assembler')
+
+        return ' '.join(cmd)
+
+
+    def run_spades_once(self, kmer, outdir):
+        cmd = self._make_spades_command(kmer, outdir)
         return common.syscall(cmd, verbose=self.verbose, allow_fail=True)
 
 

circlator/common.py

@@ -4,7 +4,7 @@
 
 class Error (Exception): pass
 
-version = '1.1.5'
+version = '1.2.0'
 
 def syscall(cmd, allow_fail=False, verbose=False):
     if verbose:

circlator/data/test_make_data.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+# This is the script that was used to generate the
+# test data, which is used when running 'circlator test'
+set -e
+set -x
+
+ref=test_ref.fa
+for_reads=test_for_reads.fa
+reads=test_reads.fq.gz
+contigs=test_contigs.fa
+
+fastaq make_random_contigs --seed 42 1 300000 $ref
+samtools faidx $ref
+
+echo ">1.twice" > $for_reads.$$
+samtools faidx $ref 1 1 | grep -v ">" >> $for_reads.$$
+fastaq to_fasta $for_reads.$$ $for_reads
+rm $for_reads.$$
+fastaq to_perfect_reads --seed 42 $for_reads $reads 16000 1 20 8000
+
+
+samtools faidx test_ref.fa 1:500-148000 > $contigs
+samtools faidx test_ref.fa 1:150000-299500 >> $contigs

circlator/data/test_ref.fa.fai

@@ -0,0 +1 @@
+1	300000	3	60	61

circlator/merge.py

@@ -24,6 +24,8 @@ def __init__(
           min_spades_circular_percent=95,
           spades_kmers=None,
           spades_use_first_success=False,
+          spades_careful=True,
+          spades_only_assembler=True,
           ref_end_tolerance=15000,
           qry_end_tolerance=1000,
           verbose=False,
@@ -45,6 +47,8 @@ def __init__(
         self.min_spades_circular_percent = min_spades_circular_percent
         self.spades_kmers = spades_kmers
         self.spades_use_first_success = spades_use_first_success
+        self.spades_careful = spades_careful
+        self.spades_only_assembler = spades_only_assembler
         self.ref_end_tolerance = ref_end_tolerance
         self.qry_end_tolerance = qry_end_tolerance
         self.verbose = verbose
@@ -698,6 +702,8 @@ def _iterative_bridged_contig_pair_merge(self, outprefix):
                     reads_prefix + '.fasta',
                     assembler_dir,
                     threads=self.threads,
+                    careful=self.spades_careful,
+                    only_assembler=self.spades_only_assembler,
                     verbose=self.verbose,
                     spades_kmers=self.spades_kmers,
                     spades_use_first_success=self.spades_use_first_success,

circlator/tasks/all.py

@@ -34,6 +34,8 @@ def run():
     assemble_group = parser.add_argument_group('assemble options')
     parser.add_argument('--assemble_spades_k', help='Comma separated list of kmers to use when running SPAdes. Max kmer is 127 and each kmer should be an odd integer [%(default)s]', default='127,117,107,97,87,77', metavar='k1,k2,k3,...')
     parser.add_argument('--assemble_spades_use_first', action='store_true', help='Use the first successful SPAdes assembly. Default is to try all kmers and use the assembly with the largest N50')
+    parser.add_argument('--assemble_not_careful', action='store_true', help='Do not use the --careful option with SPAdes (used by default)')
+    parser.add_argument('--assemble_not_only_assembler', action='store_true', help='Do not use the --assemble-only option with SPAdes (used by default)')
 
     merge_group = parser.add_argument_group('merge options')
     merge_group.add_argument('--merge_diagdiff', type=int, help='Nucmer diagdiff option [%(default)s]', metavar='INT', default=25)
@@ -143,6 +145,8 @@ def run():
         filtered_reads,
         assembly_dir,
         threads=options.threads,
+        careful=not options.assemble_not_careful,
+        only_assembler=not options.assemble_not_only_assembler,
         spades_kmers=options.assemble_spades_k,
         spades_use_first_success=options.assemble_spades_use_first,
         verbose=options.verbose
@@ -177,6 +181,8 @@ def run():
         min_spades_circular_percent=options.merge_min_spades_circ_pc,
         spades_kmers=options.assemble_spades_k,
         spades_use_first_success=options.assemble_spades_use_first,
+        spades_careful=not options.assemble_not_careful,
+        spades_only_assembler=not options.assemble_not_only_assembler,
         nucmer_breaklen=options.merge_breaklen,
         ref_end_tolerance=options.merge_ref_end,
         qry_end_tolerance=options.merge_reassemble_end,

circlator/tasks/assemble.py

@@ -7,6 +7,8 @@ def run():
     parser = argparse.ArgumentParser(
         description = 'Assemble reads using SPAdes',
         usage = 'circlator assemble [options] <in.reads.fasta> <out_dir>')
+    parser.add_argument('--not_careful', action='store_true', help='Do not use the --careful option with SPAdes (used by default)')
+    parser.add_argument('--not_only_assembler', action='store_true', help='Do not use the --assemble-only option with SPAdes (used by default)')
     parser.add_argument('--threads', type=int, help='Number of threads [%(default)s]', default=1, metavar='INT')
     parser.add_argument('--verbose', action='store_true', help='Be verbose')
     parser.add_argument('--spades_k', help='Comma separated list of kmers to use when running SPAdes. Max kmer is 127 and each kmer should be an odd integer [%(default)s]', default='127,117,107,97,87,77', metavar='k1,k2,k3,...')
@@ -19,6 +21,8 @@ def run():
         options.reads,
         options.out_dir,
         threads=options.threads,
+        careful=not options.not_careful,
+        only_assembler=not options.not_only_assembler,
         spades_kmers=options.spades_k,
         spades_use_first_success=options.spades_use_first,
         verbose=options.verbose

circlator/tasks/merge.py

@@ -13,6 +13,8 @@ def run():
     parser.add_argument('--min_length_merge', type=int, help='Minimum length of nucmer hit to use when merging [%(default)s]', metavar='INT', default=4000)
     parser.add_argument('--breaklen', type=int, help='breaklen option used by nucmer [%(default)s]', metavar='INT', default=500)
     parser.add_argument('--min_spades_circ_pc', type=float, help='Min percent of contigs needed to be covered by nucmer hits to spades circular contigs [%(default)s]', metavar='FLOAT', default=95)
+    parser.add_argument('--assemble_not_careful', action='store_true', help='Do not use the --careful option with SPAdes (used by default)')
+    parser.add_argument('--assemble_not_only_assembler', action='store_true', help='Do not use the --assemble-only option with SPAdes (used by default)')
     parser.add_argument('--spades_k', help='Comma separated list of kmers to use when running SPAdes. Max kmer is 127 and each kmer should be an odd integer [%(default)s]', default='127,117,107,97,87,77', metavar='k1,k2,k3,...')
     parser.add_argument('--spades_use_first', action='store_true', help='Use the first successful SPAdes assembly. Default is to try all kmers and use the assembly with the largest N50')
     parser.add_argument('--ref_end', type=int, help='max distance allowed between nucmer hit and end of input assembly contig [%(default)s]', metavar='INT', default=15000)
@@ -35,6 +37,8 @@ def run():
         nucmer_min_length_for_merges=options.min_length_merge,
         nucmer_breaklen=options.breaklen,
         min_spades_circular_percent=options.min_spades_circ_pc,
+        spades_careful=not options.assemble_not_careful,
+        spades_only_assembler=not options.assemble_not_only_assembler,
         spades_kmers=options.spades_k,
         spades_use_first_success=options.spades_use_first,
         ref_end_tolerance=options.ref_end,

circlator/tasks/test.py

@@ -0,0 +1,55 @@
+import argparse
+import os
+import shutil
+import sys
+import subprocess
+import circlator
+
+def run():
+    parser = argparse.ArgumentParser(
+        description = 'Run Circlator on a small test dataset',
+        usage = 'ariba test [options] <outdir>')
+    parser.add_argument('--threads', type=int, help='Number of threads [%(default)s]', default=1, metavar='INT')
+    parser.add_argument('outdir', help='Name of output directory')
+    options = parser.parse_args()
+
+    print('Running Circlator on test data...')
+
+    try:
+        os.mkdir(options.outdir)
+        os.chdir(options.outdir)
+    except:
+        print('Error making output directory "', options.outdir, '". Cannot continue.', sep='', file=sys.stderr)
+        sys.exit(1)
+
+    print('Made output directory. Copying test data files into it:')
+
+    modules_dir = os.path.dirname(os.path.abspath(circlator.__file__))
+    test_data_dir = os.path.join(modules_dir, 'data')
+
+    for filename in ['test_contigs.fa', 'test_reads.fq.gz']:
+        shutil.copy(os.path.join(test_data_dir, filename), filename)
+        print('    copied', filename)
+
+
+    cmd = ' '.join([
+        sys.argv[0],
+        'all',
+        '--threads', str(options.threads),
+        '--verbose',
+        '--assemble_spades_use_first',
+        'test_contigs.fa',
+        'test_reads.fq.gz',
+        'OUT',
+    ])
+
+    print('\nRunning Circlator with:', cmd, '', sep='\n')
+
+    return_code = subprocess.call(cmd, shell=True)
+
+    if return_code != 0:
+        print('\nSomething went wrong. See above for error message(s). Return code was', return_code)
+        sys.exit(1)
+
+    print('-' * 79)
+    print('Finished run on test data OK')

circlator/tests/assemble_test.py

@@ -38,3 +38,26 @@ def test_build_spades_kmers(self):
         with self.assertRaises(assemble.Error):
             self.assembler._build_spades_kmers('41,spam')
 
+
+    def test_make_spades_command(self):
+        '''test _make_spades_command'''
+        cmd_start = ' '.join([
+            self.assembler.spades.exe(),
+            '-s', os.path.join(data_dir, 'assemble_test.dummy_reads.fa'),
+        ])
+
+        self.assertEqual(cmd_start + ' -o out -t 1 -k 41 --careful --only-assembler', self.assembler._make_spades_command(41, 'out'))
+        self.assertEqual(cmd_start + ' -o out -t 1 -k 43 --careful --only-assembler', self.assembler._make_spades_command(43, 'out'))
+        self.assertEqual(cmd_start + ' -o out2 -t 1 -k 41 --careful --only-assembler', self.assembler._make_spades_command(41, 'out2'))
+
+        self.assembler.careful = False
+        self.assertEqual(cmd_start + ' -o out -t 1 -k 41 --only-assembler', self.assembler._make_spades_command(41, 'out'))
+        self.assembler.careful = True
+
+        self.assembler.only_assembler = False
+        self.assertEqual(cmd_start + ' -o out -t 1 -k 41 --careful', self.assembler._make_spades_command(41, 'out'))
+        self.assembler.only_assembler = True
+
+        self.assembler.threads = 2
+        self.assertEqual(cmd_start + ' -o out -t 2 -k 41 --careful --only-assembler', self.assembler._make_spades_command(41, 'out'))
+

scripts/circlator

@@ -14,6 +14,7 @@ tasks = {
     'fixstart': 'Change start position of circular sequences',
     'minimus2': 'Run the minimus2 based circularisation pipeline',
     'get_dnaa': 'Download file of dnaA (or other of user\'s choice) genes',
+    'test': 'Run Circlator on a small test set',
     'version': 'Print version and exit',
 }
 
@@ -29,6 +30,7 @@ ordered_tasks = [
     'minimus2',
     'get_dnaa',
     'progcheck',
+    'test',
     'version'
 ]
 

setup.py

@@ -7,7 +7,7 @@
 
 setup(
     name='circlator',
-    version='1.1.5',
+    version='1.2.0',
     description='circlator: a tool to circularise genome assemblies',
     packages = find_packages(),
     package_data={'circlator': ['data/*']},