v1.01 added pfam2go; split nogsql into smaller files to not have issu…

…es with git lfs; other minor improvements in stability;

Former-commit-id: dc8bc26
Former-commit-id: d636297

Resources/NOG_sql/NOGT1.tar.gz.REMOVED.git-id

@@ -1 +1 @@
-670cbc527fc2bd39173cde0284547bd5cbaff07a
+7d083d9f3955b1fcc89803f81b5562d6d0237ff4

Resources/NOG_sql/NOGT2.tar.gz.REMOVED.git-id

@@ -1 +1 @@
-a91511c3ea1e10a3ee7e19600185f81971e1ffbd
+5dcaeb873b827d6a9cf12d6f356367aaa1403818

Resources/NOG_sql/NOGT3.tar.gz.REMOVED.git-id

@@ -1 +1 @@
-3b1f9cc98e29c6488969ffbe4f39ca08ddd29f50
+e95cb23c6c2205c1b3101267e0ea2d966b18bae2

Resources/NOG_sql/NOGT4.tar.gz.REMOVED.git-id

@@ -1 +1 @@
-c226af7eae624cf7fc914db2c202e49a0493a5b7
+95a1174d032def37115f493d14434adc00847fca

Resources/NOG_sql/NOGT5.tar.gz.REMOVED.git-id

@@ -1 +1 @@
-5e0d561aadc95ea5542141397553d863a8af22b6
+1f7add08592b207f63f09a1b1bc883e87da11e29

source/Exceptions.py

@@ -21,7 +21,7 @@ def __str__(self):
 
 class InvalidTargetFile(Exception):
     def __str__(self):
-        return 'You did not insert a valid target file! Make sure it follows the same format as the file provided in tests/test_file.tsv'
+        return 'You did not insert a valid target file!'
 
 class InstallationCheckNotPassed(Exception):
     def __str__(self):

source/MANTIS.py

@@ -148,6 +148,7 @@ def __str__(self):
         'User cores:\t\t\t\t'+str(self.user_cores)+'\n' if self.user_cores else '',
         'Chunk size:\t\t\t'+str(self.chunk_size)+'\n' if self.chunk_size else '',
         'Domain algorithm:\t\t'+str(self.domain_algorithm)+'\n' if self.domain_algorithm else '',
+        'Skip consensus:\t\t'+str(self.skip_consensus)+'\n' if self.skip_consensus else '',
         '------------------------------------------']
         return 'User configuration:'+'\n'+'------------------------------------------'+'\n'+''.join(output_list)
 
@@ -165,14 +166,16 @@ def generate_fastas_to_annotate(self):
         else:
             print('Your file does not appear to be a fasta. If you want to annotate multiple samples, make sure your file has the <.tsv> extension.', flush=True, file=self.redirect_verbose)
             raise InvalidTargetFile
+        if not self.fastas_to_annotate: raise InvalidTargetFile
         for file_path, output_path, organism_details, count_seqs_original_file in self.fastas_to_annotate:
             Path(output_path).mkdir(parents=True, exist_ok=True)
 
     def annotate_multiple_samples(self):
         try:
             with open(self.target_path) as file:
                 line = file.readline()
-                line = file.readline()
+                if splitter not in line:
+                    line = file.readline()
                 while line:
                     line = line.strip('\n').split()
                     if len(line) >= 2:

source/MANTIS_Assembler.py

@@ -310,10 +310,7 @@ def merge_hmm_folder(self, target_folder):
         run_command('for i in ' + target_folder + '*.hmm; do cat ${i} >> ' + target_folder + output_file + '_merged.hmm; done',stdout_file=self.redirect_verbose)
         run_command('hmmpress '+target_folder + output_file + '_merged.hmm',stdout_file=self.redirect_verbose)
 
-    def file_exists(self,target_file,force_download=False):
-        if os.path.exists(target_file) and not force_download:
-            return True
-        return False
+
 
     def get_path_default_hmm(self,database,taxon_id=None):
         target_file=None
@@ -339,21 +336,21 @@ def get_path_default_hmm(self,database,taxon_id=None):
 
     def check_reference_exists(self,database,taxon_id=None,force_download=False):
         if database=='go_obo_nlp':
-            if self.file_exists(self.mantis_paths['go_obo_nlp'] + 'go.obo', force_download):
+            if file_exists(self.mantis_paths['go_obo_nlp'] + 'go.obo', force_download):
                 return True
         elif database=='uniprot_nlp':
             if os.listdir(self.mantis_paths['uniprot_nlp']):
                 return True
         elif database=='ncbi':
-            if self.file_exists(self.mantis_paths['ncbi'] + 'taxidlineage.dmp', force_download):
+            if file_exists(self.mantis_paths['ncbi'] + 'taxidlineage.dmp', force_download):
                 return True
         elif database=='NOGSQL':
-            if self.file_exists(self.mantis_paths['default'] + 'eggnog.db', force_download):
+            if file_exists(self.mantis_paths['default'] + 'eggnog.db', force_download):
                 return True
         target_file=self.get_path_default_hmm(database,taxon_id)
         if target_file:
             for extension in ['','.h3f','.h3i','.h3m','.h3p']:
-                if not self.file_exists(target_file+extension,force_download=force_download):
+                if not file_exists(target_file+extension,force_download=force_download):
                     return False
         else:
             return False
@@ -363,15 +360,15 @@ def check_reference_exists(self,database,taxon_id=None,force_download=False):
 
     def check_installation_extras(self,res,verbose=True):
         if verbose: yellow('Checking extra files',flush=True,file=self.redirect_verbose)
-        if not self.file_exists(self.mantis_paths['resources'] + 'essential_genes/essential_genes.txt'):
+        if not file_exists(self.mantis_paths['resources'] + 'essential_genes/essential_genes.txt'):
             red('Essential genes list is missing, it should be in the github repo!')
             if verbose: red('Failed installation check on [files missing]: ' + self.mantis_paths['resources'] + 'essential_genes/essential_genes.txt', flush=True,file=self.redirect_verbose)
             res.append(self.mantis_paths['resources'] + 'essential_genes/')
         else:
             if verbose: green('Passed installation check on: ' + self.mantis_paths['resources'] + 'essential_genes', flush=True,file=self.redirect_verbose)
 
 
-        if not self.file_exists(self.mantis_paths['ncbi'] + 'taxidlineage.dmp'):
+        if not file_exists(self.mantis_paths['ncbi'] + 'taxidlineage.dmp'):
             if verbose: red('Failed installation check on [files missing]: ' + self.mantis_paths['ncbi']+'taxidlineage.dmp', flush=True, file=self.redirect_verbose)
             res.append(self.mantis_paths['ncbi'])
         else:
@@ -387,7 +384,7 @@ def check_installation_extras(self,res,verbose=True):
             else:
                 if verbose: green('Passed installation check on: ' + self.mantis_paths['uniprot_nlp'], flush=True, file=self.redirect_verbose)
 
-        if not self.file_exists(self.mantis_paths['go_obo_nlp'] + 'go.obo'):
+        if not file_exists(self.mantis_paths['go_obo_nlp'] + 'go.obo'):
             if verbose: red('Failed installation check on [files missing]: ' + self.mantis_paths['go_obo_nlp']+'go.obo', flush=True, file=self.redirect_verbose)
             res.append(self.mantis_paths['go_obo_nlp'])
         else:
@@ -449,7 +446,7 @@ def check_installation(self,verbose=True):
         if verbose: yellow('Checking HMM installation',flush=True,file=self.redirect_verbose)
         requirements={
             self.mantis_paths['NOGG']:['NOGG_sql_annotations.tsv'],
-            self.mantis_paths['pfam']:['Pfam-A.hmm.dat'],
+            self.mantis_paths['pfam']:['pfam_metadata.tsv'],
             self.mantis_paths['kofam']:['ko_list','ko2cog.xl','ko2go.xl','ko2tc.xl','ko2cazy.xl','ko_to_path','map_description'],
             #self.mantis_paths['dbcan']:['CAZyDB.07312019.fam.subfam.ec.txt'],
             self.mantis_paths['tigrfam']:['gpl.html','COPYRIGHT','TIGRFAMS_GO_LINK','TIGRFAMS_ROLE_LINK','TIGR_ROLE_NAMES'],
@@ -583,4 +580,5 @@ def processes_handler(self,target_worker_function,worker_count,add_sentinels=Tru
 
 
 if __name__ == '__main__':
-    p=MANTIS_Assembler()
+    p=MANTIS_Assembler()
+    p.unpack_NOG_sql()