Fix Dicey, using Python 3, NamedTemporaryFile, new index paths

Author: kevinlul

Diff for: .gitignore

@@ -1,8 +1,5 @@
 # program/data folders
 .env
-bin/dicey
-bin/ncbi-blast-2.7.1+
-bin/primer3
 jbrowse/data*
 config
 mongodb

Diff for: config-template/paths.conf

@@ -1,5 +1,3 @@
 PRIMER3_EXEC=/usr/bin/primer3_core
 PRIMER3_CONFIG=/etc/primer3_config
 BLAST_EXEC=/usr/bin/blastn
-DICEY_EXEC=/opt/dicey/bin/dicey
-DICEY_PATH=/opt/dicey

Diff for: docker-compose.yaml

@@ -22,4 +22,3 @@ volumes:
 # After startup, exec into the container and run
 # cd /var/www/html/src/setup
 # python3 setup.py -r /var/www/html -g Saccharomyces_cerevisiae -v R64-1-1 -fa2bit /var/www/html/bin/faToTwoBit -b /usr/bin
-# ./enable_dicey.sh Saccharomyces_cerevisiae R64-1-1

Diff for: src/helpers/Config.py

@@ -14,28 +14,28 @@
 from pymongo import MongoClient
 
 
-class Config:	
-	def __init__(self, genome=None):	
+class Config:
+	def __init__(self, genome=None):
 		# get the credentials
 		self.config = getCredentials()
 		# the root directory is 2 levels up, get its dirname
 		self.ROOT_PATH = (os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))))
 		# set all the paths defined in ROOT/config/paths.conf
-		self.setPaths()	
+		self.setPaths()
 		self.PRIMER3_SETTINGS = os.path.join(self.ROOT_PATH, 'config/primer3settings.conf')
 		# where the primer3 in and output files are written
 		self.PRIMER3_DIR = os.path.join(self.ROOT_PATH, 'src/primer-design/files/primer3Files')
-			
+
 		if genome:
 			# connect to the mongoDB
-			self.genome = genome	
+			self.genome = genome
 			mongoDB, release, curr_geneCollection, guideCollection, primerCollection, metadataCollection = self.getAttributes()
 			self.mongoDB = mongoDB
 			self.release = release
 			self.curr_geneCollection = curr_geneCollection
 			self.guideCollection = guideCollection
 			self.primerCollection = primerCollection
-			self.metadataCollection = metadataCollection	
+			self.metadataCollection = metadataCollection
 			self.organismName = self.getOrgName()
 
 
@@ -48,24 +48,13 @@ def setPaths(self):
 				self.PRIMER3 = line.split('=')[1].strip()
 			elif re.match(r"^PRIMER3_CONFIG=", line):
 				self.PRIMER3_CONFIG = line.split('=')[1].strip()
-			elif re.match(r"DICEY_EXEC=", line):
-				self.DICEY = line.split('=')[1].strip()
-			elif re.match(r"DICEY_PATH=", line):
-				self.DICEY_PATH = line.split('=')[1].strip()
 
 		if not self.BLAST:
 			print("Error: path to BLAST executable not defined in paths.conf")
 		elif not self.PRIMER3:
 			print("Error: path to primer3 executable not defined in paths.conf")
 		elif not self.PRIMER3_CONFIG:
 			print("Error: path to primer3 config directory not defined in paths.conf")
-		elif not self.DICEY:
-			print("Error: path to dicey executable not defined in paths.conf")
-		elif not self.DICEY_PATH:
-			print("Error: path to dicey program folder not defined in paths.conf")
-
-		return
-				
 
 	def getAttributes(self):
 		if self.config:
@@ -74,10 +63,10 @@ def getAttributes(self):
 		else:
 			# otherwise, attempt connecting without credentials
 			client = MongoClient()
-		
+
 		db = client[self.genome]
 		collections = db.collection_names()
-		
+
 		# get the release we're using from the most recent geneInfo collection
 		curr_release = fetchCurrentRelease(self.genome, collections)
 
@@ -86,23 +75,23 @@ def getAttributes(self):
 		nameGuideCol = "gRNAResultCollection"
 		namePrimerCol = "primerCollection"
 		metadataCol = "metadata"
-				
+
 		return db, curr_release, db[nameGeneCol], db[nameGuideCol], db[namePrimerCol], db[metadataCol]
 
 
 	def getOrgName(self):
-		orgName = ''	
+		orgName = ''
 		metadataRecords = self.metadataCollection.find({})
 		if metadataRecords.count() == 1:
-			for record in metadataRecords:	
-				orgName = str(record['org_name'])	
+			for record in metadataRecords:
+				orgName = str(record['org_name'])
 		else:
 			print("Error: metadata collection for " + str(self.genome) + " is misconfigured")
-		
+
 		return orgName
-		
 
-	@staticmethod		
+
+	@staticmethod
 	def fetchStrandofGene(ensid, genome):
 		curr_geneCollection = getCurrentGeneCollection(genome)
 		record = curr_geneCollection.find_one({'ENSID': ensid})
@@ -119,40 +108,40 @@ def getCredentials():
 	except Exception, e:
 		# this is ok it just means there are no credentials
 		return
-	for line in config_file:		
+	for line in config_file:
 		var, value = line.rstrip().split("=")
 		if var == "password":
 			# passwords need to have special characters escaped
 			config[var] = urllib.quote(value)
 		else:
 			config[var] = value
-	
-	return config		
+
+	return config
 
 
 def getCurrentGeneCollection(genome):
 	# get credentials and connect to mongodb
-	config = getCredentials()	
+	config = getCredentials()
 	if config:
 		client = MongoClient('mongodb://%s:%s@localhost' % (config['username'], config['password']))
 	else:
 		# if no credentials defined
 		client = MongoClient()
-	
+
 	db = client[genome]
 	collections = db.collection_names()
-	
+
 	# get the release we're using from the most recent geneInfo collection
 	curr_release = fetchCurrentRelease(genome, collections)
 
 	# determine the collection name
 	nameGeneCol = "geneInfo_" + str(curr_release)
-	
+
 	return db[nameGeneCol]
 
 
 def fetchCurrentRelease(genome, collections=None):
-	
+
 	# optionally, if function calling already has list of connections, they can be passed
 	# otherwise, connect to mongodb and get them
 	if not collections:
@@ -171,7 +160,7 @@ def fetchCurrentRelease(genome, collections=None):
 			cltn_release = c.split("_",1)[1]
 			if int(cltn_release) > int(curr_release):
 				curr_release = cltn_release
-	
+
 	return curr_release
 
 
@@ -183,11 +172,11 @@ def fetchInstalledGenomes():
 	else:
 		client = MongoClient()
 
-	genomes = []	
+	genomes = []
 	dbs = sorted(client.database_names(), key=lambda v: v.upper())
 	default_dbs = ['admin', 'local','config']
 	for db in dbs:
-		if db not in default_dbs:	
+		if db not in default_dbs:
 			metadataCollection = client[db]['metadata']
 			metadataRecords = metadataCollection.find({})
 			if metadataRecords.count() == 1:
@@ -196,7 +185,7 @@ def fetchInstalledGenomes():
 				genomes.append((db, orgName)) # store org code and name in list as tuple
 			else:
 				print("Error: metadata collection for " + str(db) + " is misconfigured")
-	return genomes		
+	return genomes
 
 
 def main():
@@ -208,11 +197,11 @@ def main():
 	print "Databases currently available:"
 	genomes = fetchInstalledGenomes()
 	for g in genomes:
-		orgName = g[1]		
+		orgName = g[1]
 		genomeString = g[0] + " (" + orgName + ")"
 		print genomeString
-			
-	
+
+
 if __name__ == "__main__":
 	main()
-	
+

Diff for: src/helpers/Config3.py

@@ -59,24 +59,13 @@ def setPaths(self):
 				self.PRIMER3 = line.split('=')[1].strip()
 			elif re.match(r"^PRIMER3_CONFIG=", line):
 				self.PRIMER3_CONFIG = line.split('=')[1].strip()
-			elif re.match(r"DICEY_EXEC=", line):
-				self.DICEY = line.split('=')[1].strip()
-			elif re.match(r"DICEY_PATH=", line):
-				self.DICEY_PATH = line.split('=')[1].strip()
 
 		if not self.BLAST:
 			print("Error: path to BLAST executable not defined in paths.conf")
 		elif not self.PRIMER3:
 			print("Error: path to primer3 executable not defined in paths.conf")
 		elif not self.PRIMER3_CONFIG:
 			print("Error: path to primer3 config directory not defined in paths.conf")
-		elif not self.DICEY:
-			print("Error: path to dicey executable not defined in paths.conf")
-		elif not self.DICEY_PATH:
-			print("Error: path to dicey program folder not defined in paths.conf")
-
-		return
-
 
 	def getRGENs(self):
 		" return the rgen collection"

Diff for: src/primer-design/classes/Dicey.py

@@ -1,102 +1,69 @@
-#!/usr/bin/python
+#!/usr/bin/env python3
 
 """
 Hillary Elrick February 4th, 2019
 
 Class definition to ease & organize access to Dicey in silico PCR tool
 """
 
-import sys
-import re
-import os
-import subprocess
-import time
-import json
+import json, os, subprocess, sys
+from tempfile import NamedTemporaryFile
+from typing import Any, Dict, List, Optional
 
 # get the global root path from the Config object
 sys.path.append("..")
-from Config import Config
+from Config3 import Config
+
 
 class Dicey:
 	"""
 	Dicey is used to run in silico PCR on a pair of primers.
-	Default temperature for the primers is 45C
 	"""
-	def __init__(self, sequences, temp='45', genome='mm10'):
-		self.Config = Config()
-		sys.path.insert(0, self.Config.DICEY_PATH)	
+	def __init__(self, sequences: List[str], temperature: int, genome: str):
 		assert len(sequences) == 2, "Exactly two primers required"
+		self.config = Config()
 		self.sequences = sequences
-		self.temp = temp # temperature
-		self.genomePath = self.Config.DICEY_PATH+"/indexes/"+genome+"/"+genome+".fa.gz"
-		self.tempfile = self.createTempFile()	
-		self.diceyCommand = self.constructDiceyCommand()	
+		self.temperature = temperature
+		self.genome_path = os.path.join(self.config.ROOT_PATH, "jbrowse", "data", genome, "processed", genome + ".fa")
 
-	
-	def constructDiceyCommand(self):
+	@property
+	def command(self) -> List[str]:
 		"""
-		Returns the Dicey command with reference to the locally installed Primer3 in addition
-		to the genome of interest and minimum temperature to consider for binding
+		Returns the Dicey command prefix with reference to the locally installed Primer3 in addition
+		to the genome of interest and minimum temperature to consider for binding.
+		The final sequences.fasta parameter is not included.
 		"""
-		# base shell command 
-		diceyCommand = [self.Config.DICEY]
-		diceyCommand.append('search')
-
-		# provide location of primer3 config directory
-		diceyCommand.append('-i')
-		diceyCommand.append(self.Config.PRIMER3_CONFIG)
-		
-		# add in the temperature
-		diceyCommand.append('-c')
-		diceyCommand.append(self.temp)
-
-		# add in the genome
-		diceyCommand.append('-g')
-		diceyCommand.append(self.genomePath)
-
-		# add in temp file
-		diceyCommand.append(self.tempfile)
-
-		return diceyCommand
-
-
-	def createTempFile(self):
-		filename = str(self.sequences[0])+"_"+str(self.sequences[1])+(time.strftime("%Y-%m-%d-%H:%M:%S"))
-		f = open(os.path.join(self.Config.DICEY_PATH, "dicey_tempfiles", filename), "w+")
-			
-		faFormat = '>leftPrimer'
-		faFormat += '\n' + str(self.sequences[0])
-		faFormat += '\n>rightPrimer'
-		faFormat += '\n' + str(self.sequences[1])	
-	
-		f.write(faFormat)
-		f.close()	
-		
-		return f.name
-
-
-	def deleteTempFile(self):
-		if os.path.exists(self.tempfile):
-			os.remove(self.tempfile)
-
-
-	def runSequences(self):		
-		diceyProcess = subprocess.Popen(self.diceyCommand, stdout=subprocess.PIPE, stdin=subprocess.PIPE)
-		(diceyOut, diceyErr) = diceyProcess.communicate()
-
-		try:
-			# decode json
-			jsonResult = json.loads(diceyOut)
-			if 'errors' in jsonResult and len(jsonResult['errors']) > 0:
-				print("Error(s) from Dicey program: ")
-				for error in jsonResult['errors']:
-					print(error['title'])
+		return [
+			"dicey",
+			"search",
+			"--config",
+			self.config.PRIMER3_CONFIG,
+			"--cutTemp",
+			self.temperature,
+			"--genome",
+			self.genome_path
+		]
+
+
+	def run(self) -> Optional[Dict[Any]]:
+		with NamedTemporaryFile(mode="w", encoding="utf-8") as sequences:
+			sequences.write(f">leftPrimer\n{self.sequences[0]}\n>rightPrimer\n{self.sequences[1]}\n")
+			sequences.flush()
+			dicey_process = subprocess.run(self.command + [sequences.name], capture_output=True)
+			# If exceptions are okay then switch to check=True
+			if dicey_process.returncode:
+				print(dicey_process.stdout)
+				print(dicey_process.stderr)
 				return
-			data = jsonResult['data']
-		except Exception, e:
-			print("Error reading Dicey results: "+str(e))
-			return
-		
-		return data
-
-	
+			else:
+				try:
+					jsonResult = json.loads(dicey_process.stdout)
+					if 'errors' in jsonResult and len(jsonResult['errors']) > 0:
+						print("Error(s) from Dicey program: ")
+						for error in jsonResult['errors']:
+							print(error['title'])
+						return
+					return jsonResult['data']
+				except Exception as e:
+					print("Error reading Dicey results: " + str(e))
+					return