attempting to debug issues with KMC stuff for #15, so creating a new …

…branch since a lot will get messed up.
dkoslicki · Mar 31, 2020 · 1af1b2d · 1af1b2d
1 parent 8ec7889
commit 1af1b2d
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 14 deletions.
diff --git a/CMash/Query.py b/CMash/Query.py
@@ -224,9 +224,9 @@ def count_training_kmers(self):
 
 		# count the training k-mers
 		if self.verbose:
-			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} -fa -ci1 -t{self.threads} {self.cmashDump} {out_path} .", shell=True)
+			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} -fa -ci0 -cs3 -t{self.threads} {self.cmashDump} {out_path} .", shell=True)
 		else:
-			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} -fa -ci1 -t{self.threads} {self.cmashDump} {out_path} .", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} -fa -ci0 -cs3 -t{self.threads} {self.cmashDump} {out_path} .", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 		if res.returncode != 0:
 			raise Exception(f"The command {res.args} failed to run and returned the returncode={res.returncode}")
 
@@ -250,9 +250,9 @@ def count_input_kmers(self):
 		#count kmers in the input sample (not the training db)
 		# TODO: will need to try with/without -fm if type_arg == '-fa'
 		if self.verbose:
-			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} {type_arg} -ci1 -t{self.threads} -fm {self.reads_path} {out_path} {self.temp_dir}", shell=True)
+			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} {type_arg} -ci0 -cs3 -t{self.threads} -fm {self.reads_path} {out_path} {self.temp_dir}", shell=True)
 		else:
-			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} {type_arg} -ci1 -t{self.threads} -fm {self.reads_path} {out_path} {self.temp_dir}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+			res = subprocess.run(f"{self.kmc} -v -k{self.ksize} {type_arg} -ci0 -cs3 -t{self.threads} -fm {self.reads_path} {out_path} {self.temp_dir}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 		if res.returncode != 0:
 			raise Exception(f"The command {res.args} failed to run and returned the returncode={res.returncode}")
 
@@ -275,9 +275,9 @@ def intersect(self):
 
 		# intersect kmers
 		if self.verbose:
-			res = subprocess.run(f"{self.kmc_tools} simple {db_kmers_loc} {in_path} intersect {out_path}", shell=True)
+			res = subprocess.run(f"{self.kmc_tools} simple {db_kmers_loc} {in_path} -ci0 intersect {out_path} -ci0 ", shell=True)
 		else:
-			res = subprocess.run(f"{self.kmc_tools} simple {db_kmers_loc} {in_path} intersect {out_path}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+			res = subprocess.run(f"{self.kmc_tools} simple {db_kmers_loc} {in_path} -ci0 intersect {out_path} -ci0 ", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 		if res.returncode != 0:
 			raise Exception(f"The command {res.args} failed to run and returned the returncode={res.returncode}")
 
@@ -286,9 +286,9 @@ def intersect(self):
 		if self.verbose:
 			print("dumping intersection to FASTA file")
 		if self.verbose:
-			res = subprocess.run(f"{self.kmc_dump} {out_path} {dump_path}", shell=True)
+			res = subprocess.run(f"{self.kmc_dump} -ci0 {out_path} {dump_path}", shell=True)
 		else:
-			res = subprocess.run(f"{self.kmc_dump} {out_path} {dump_path}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+			res = subprocess.run(f"{self.kmc_dump} -ci0 {out_path} {dump_path}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 		if res.returncode != 0:
 			raise Exception(f"The command {res.args} failed to run and returned the returncode={res.returncode}")
 

diff --git a/scripts/StreamingQueryDNADatabase.py b/scripts/StreamingQueryDNADatabase.py
@@ -190,7 +190,7 @@ def map_func(sequence):
 	if args.intersect:
 		intersecter = Intersect(query_file, training_database_file)
 		intersecter.compute_intersection()
-		print("intersection located at: {}".format(intersecter.out_file))
+		#print("intersection located at: {}".format(intersecter.out_file))
 		# change query file to the intersection file
 		query_file = intersecter.out_file
 	# Open the file to prepare for processing

diff --git a/tests/script_tests/run_small_tests.sh b/tests/script_tests/run_small_tests.sh
@@ -41,7 +41,7 @@ fi
 echo "Classifying sample, sensitive settings"
 rm results.csv 2> /dev/null
 # make a streaming pre-filter
-/usr/bin/time python ${scriptsDir}/StreamingQueryDNADatabase.py ${testOrganism} TrainingDatabase.h5 results.csv 10-21-2 --sensitive
+/usr/bin/time python ${scriptsDir}/StreamingQueryDNADatabase.py ${testOrganism} TrainingDatabase.h5 results.csv 10-21-1 --sensitive
 if test -f results.csv; then
   echo "sensitive classify successful"
   cat results.csv
@@ -50,7 +50,7 @@ else
   exit 1
 fi
 
-
+: << 'END1'
 echo "Classifying sample, specific settings"
 rm results.csv 2> /dev/null
 # make a streaming pre-filter
@@ -62,13 +62,14 @@ else
   echo "SOMETHING WENT WRONG!!!!"
   exit 1
 fi
+END1
 
 # intersection tests
 
 echo "Classifying sample, sensitive settings, with KMC intersect"
 rm results.csv 2> /dev/null
 # make a streaming pre-filter
-/usr/bin/time python ${scriptsDir}/StreamingQueryDNADatabase.py ${testOrganism} TrainingDatabase.h5 results.csv 10-21-2 --sensitive --intersect
+/usr/bin/time python ${scriptsDir}/StreamingQueryDNADatabase.py ${testOrganism} TrainingDatabase.h5 results.csv 10-21-1 --sensitive --intersect
 if test -f results.csv; then
   echo "sensitive classify successful"
   cat results.csv
@@ -77,7 +78,7 @@ else
   exit 1
 fi
 
-
+: << 'END'
 echo "Classifying sample, specific settings, with KMC intersect"
 rm results.csv 2> /dev/null
 # make a streaming pre-filter
@@ -88,4 +89,5 @@ if test -f results.csv; then
 else
   echo "SOMETHING WENT WRONG!!!!"
   exit 1
-fi
+fi
+END