Merge pull request #2282 from merenlab/prodigal-segmentation-fault-fixy

A solution for prodigal segmentation fault errors

anvio/__init__.py

@@ -310,6 +310,19 @@ def TABULATE(table, header, numalign="right", max_width=0):
                      "frames in contigs by running a bacterial gene caller. Declaring this flag will by-pass that "
                      "process. If you prefer, you can later import your own gene calling results into the database."}
                 ),
+    'prodigal-single-mode': (
+            ['--prodigal-single-mode'],
+            {'default': False,
+             'action': 'store_true',
+             'help': "By default, anvi'o will use prodigal for gene calling (unless you skipped gene calling, or provided "
+                     "anvi'o with external gene calls). One of the flags anvi'o includes in prodigal run is `-p meta`, which "
+                     "optimizes prodigal's ability to identify genes in metagenomic assemblies. In some rare cases, for a "
+                     "given set of contigs prodigal will yield a segmentation fault error due to one or more genes in your "
+                     "collections will confuse the program when it is used with the `-p meta` flag. While anvi'o developers "
+                     "are not quite sure under what circumstances this happens, we realized that removal of this flag often "
+                     "solves this issue. If you are dealing with such cyrptic errors, the inclusion of `--skip-prodigal-meta-flag` "
+                     "will instruct anvi'o to run prodigal without the `-meta` flag, and may resolve this issue for you."}
+                ),
     'remove-partial-hits': (
             ['--remove-partial-hits'],
             {'default': False,

anvio/dbops.py

@@ -4316,6 +4316,7 @@ def create(self, args):
         ignore_internal_stop_codons = A('ignore_internal_stop_codons')
         skip_predict_frame= A('skip_predict_frame')
         prodigal_translation_table = A('prodigal_translation_table')
+        prodigal_single_mode = A('prodigal_single_mode')
 
         if external_gene_calls_file_path:
             filesnpaths.is_proper_external_gene_calls_file(external_gene_calls_file_path)
@@ -4516,6 +4517,8 @@ def create(self, args):
         if external_gene_calls_file_path:
             self.run.info('External gene calls file have AA sequences?', external_gene_calls_include_amino_acid_sequences, mc='green')
             self.run.info('Proper frames will be predicted?', (not skip_predict_frame), mc='green')
+        else:
+            self.run.info('Is prodigal run in single mode?', ('YES' if prodigal_single_mode else 'NO'), mc='green')
 
         self.run.info('Ignoring internal stop codons?', ignore_internal_stop_codons)
         self.run.info('Splitting pays attention to gene calls?', (not skip_mindful_splitting))

anvio/drivers/prodigal.py

@@ -35,10 +35,10 @@ def __init__(self, args=None, progress=progress, run=run):
         self.progress = progress
         self.run = run
         self.args = args
-
         A = lambda x: (args.__dict__[x] if x in args.__dict__ else None) if args else None
         self.prodigal_translation_table = A('prodigal_translation_table')
         self.num_threads = A('num_threads')
+        self.prodigal_single_mode = A('prodigal_single_mode')
 
         self.run.info('Num threads for gene calling', self.num_threads)
 
@@ -89,6 +89,7 @@ def __parser_1(self, defline):
 
         return hit
 
+
     def check_version(self):
         """checks the installed version of prodigal, sets the parser"""
 
@@ -105,6 +106,7 @@ def check_version(self):
         self.installed_version = version_found
         self.parser = self.available_parsers[version_found]
 
+
     def process(self, fasta_file_path, output_dir):
         """Take the fasta file, run prodigal on it, and make sense of the output
 
@@ -116,16 +118,17 @@ def process(self, fasta_file_path, output_dir):
         self.genes_in_contigs = os.path.join(output_dir, 'contigs.genes')
         self.amino_acid_sequences_in_contigs = os.path.join(output_dir, 'contigs.amino_acid_sequences')
 
+        self.run.warning("Anvi'o will use 'prodigal' by Hyatt et al (doi:10.1186/1471-2105-11-119) to identify open "
+                         "reading frames in your data. When you publish your findings, please do not forget to "
+                         "properly credit their work.", lc='green', header="CITATION")
+
         # Put some nice logging info.
-        self.run.warning('', header='Finding ORFs in contigs', lc='green')
+        self.run.warning('', header='Finding ORFs in contigs using prodigal', lc='green')
+        self.run.info('Procedure', 'single' if self.prodigal_single_mode else 'meta')
         self.run.info('Genes', self.genes_in_contigs)
         self.run.info('Amino acid sequences', self.amino_acid_sequences_in_contigs)
         self.run.info('Log file', log_file_path)
 
-        self.run.warning("Anvi'o will use 'prodigal' by Hyatt et al (doi:10.1186/1471-2105-11-119) to identify open "
-                         "reading frames in your data. When you publish your findings, please do not forget to "
-                         "properly credit their work.", lc='green', header="CITATION")
-
         self.progress.new('Processing')
         self.progress.update(f"Identifying ORFs using {terminal.pluralize('thread', self.num_threads)}.")
 
@@ -153,7 +156,8 @@ def process(self, fasta_file_path, output_dir):
                                   logger=terminal.Logger(progress=self.progress, run=self.run),
                                   installed_version=self.installed_version,
                                   parser=self.parser,
-                                  translation_table=self.prodigal_translation_table)
+                                  translation_table=self.prodigal_translation_table,
+                                  prodigal_single_mode=self.prodigal_single_mode)
 
         prodigal_runner = ThreadedProdigalRunner(args)
 

anvio/tables/genecalls.py

@@ -484,7 +484,7 @@ def run_gene_caller(self, gene_caller='prodigal'):
 
         if not self.contigs_fasta:
             self.contigs_fasta = filesnpaths.get_temp_file_path()
-            utils.export_sequences_from_contigs_db(self.contigs_db_path,
+            utils.export_sequences_from_contigs_db(self.db_path,
                                                    output_file_path=self.contigs_fasta,
                                                    run=self.run)
             remove_fasta_after_processing = True
@@ -495,7 +495,24 @@ def run_gene_caller(self, gene_caller='prodigal'):
 
         gene_caller = genecalling.GeneCaller(self.contigs_fasta, gene_caller=gene_caller, args=self.args, debug=self.debug)
 
-        gene_calls_dict, amino_acid_sequences = gene_caller.process()
+        try:
+            gene_calls_dict, amino_acid_sequences = gene_caller.process()
+        except Exception as e:
+            if 'prodigal' in e.e:
+                self.run.warning("There was a problem with your gene calling, and the error seems to be related to 'prodigal'. Please "
+                                 "find additional details below. It is difficult to determine what caused this error, but if you would "
+                                 "like to be certain, you can literally copy the command shown below into a single line, and run on "
+                                 "the same machine manually (or re-run the same command you run to get this error with the `--debug` flag "
+                                 "to keep all the original log files from profigal). If this error is due to a 'segmentation fault', "
+                                 "please consider including `--prodigal-single-mode` flag `anvi-gen-contigs-database` command. More "
+                                 "information `--prodigal-single-mode` is available in the help menu of `anvi-gen-contigs-database`.",
+                                 header="💀 PRODIGAL FAILED 💀")
+
+            # remove the unfinished contigs-db file
+            os.remove(self.db_path)
+
+            # show the user the actual error from down below
+            raise ConfigError(f"{e}")
 
         if not self.debug and remove_fasta_after_processing:
             os.remove(self.contigs_fasta)

anvio/threadingops.py

@@ -357,7 +357,7 @@ def __init__(self, args):
         A = lambda x: args.__dict__[x] if x in args.__dict__ else None
 
         required_args = ['input_file_path', 'collated_output_file_paths', 'number_of_splits', 'log_file_path',
-                         'installed_version', 'parser']
+                         'installed_version', 'parser', 'prodigal_single_mode']
 
         # Check that the required arguments are present.
         for arg in required_args:
@@ -370,6 +370,7 @@ def __init__(self, args):
                          log_file_path=A('log_file_path'),
                          logger=A('logger'))
 
+        self.prodigal_single_mode = A('prodigal_single_mode')
         self.installed_version = A('installed_version')
         self.parser = A('parser')
 
@@ -380,6 +381,7 @@ def __init__(self, args):
 
         self.translation_table = translation_table
 
+
     # Implement the abstract methods
     #
     #
@@ -401,6 +403,7 @@ def _split_input_file(self):
 
         return State(input_file_splits=self.input_file_splits)
 
+
     def _make_commands(self):
         """Make commands and store them in `self.commands`.
 
@@ -445,8 +448,14 @@ def _make_commands(self):
                         "you are reading this message, then please contact an anvi'o developer."
                         % str(self.translation_table))
             else:
-                # Use 'meta' mode if no translation tables are given.
-                command.extend(['-p', 'meta'])
+                if self.prodigal_single_mode:
+                    # the user explicitly requested to not use the `-p meta` flag to run
+                    # prodigal (the default procedure is single, so prodigal will fall back
+                    # to 'single' mode in this case)
+                    pass
+                else:
+                    # Use 'meta' mode if no translation tables are given.
+                    command.extend(['-p', 'meta'])
 
             self.commands.append(command)
 

bin/anvi-gen-contigs-database

@@ -53,6 +53,7 @@ if __name__ == '__main__':
 
     groupD = parser.add_argument_group('GENES IN CONTIGS', 'Expert thingies.')
     groupD.add_argument(*anvio.A('skip-gene-calling'), **anvio.K('skip-gene-calling'))
+    groupD.add_argument(*anvio.A('prodigal-single-mode'), **anvio.K('prodigal-single-mode'))
     groupD.add_argument(*anvio.A('prodigal-translation-table'), **anvio.K('prodigal-translation-table'))
     groupD.add_argument(*anvio.A('external-gene-calls'), **anvio.K('external-gene-calls'))
     groupD.add_argument(*anvio.A('ignore-internal-stop-codons'), **anvio.K('ignore-internal-stop-codons'))