Directly download UCSC Genome Browser kent binaries to /usr/local/bin

Author: kevinlul

Diff for: Dockerfile

@@ -32,6 +32,9 @@ RUN curl -LO https://jbrowse.org/releases/JBrowse-${JBROWSE_VERSION}/JBrowse-${J
     cd /var/www/html/jbrowse && ./setup.sh
 RUN pip install --no-cache-dir pymongo==3.8.0 requests==2.22.0 && \
     pip3 install --no-cache-dir pymongo==3.8.0 requests==2.20.0 Jinja2==3.1.2
+RUN curl -Lo /usr/local/bin/faToTwoBit https://hgdownload.soe.ucsc.edu/admin/exe/linux.x86_64/faToTwoBit \
+         -Lo /usr/local/bin/twoBitToFa https://hgdownload.soe.ucsc.edu/admin/exe/linux.x86_64/twoBitToFa && \
+    chmod +x /usr/local/bin/faToTwoBit /usr/local/bin/twoBitToFa
 COPY config-template /var/www/html/config
 WORKDIR /var/www/html
 CMD service mongodb start && exec apache2-foreground

Diff for: bin/faToTwoBit

@@ -6,11 +6,12 @@ services:
     ports:
       - 80:80
     tmpfs:
-      - /run
+      - /run/apache2
+      - /run/mongodb
+      - /run/lock
       - /tmp
     volumes:
       - ./sites-enabled:/etc/apache2/sites-enabled
-      - ./bin:/var/www/html/bin
       - ./docs:/var/www/html/docs
       - ./src:/var/www/html/src
       - ./jbrowse/data:/var/www/html/jbrowse/data

Diff for: bin/twoBitToFa

@@ -1,4 +1,4 @@
-#!/usr/bin/python3
+#!/usr/bin/env python3
 
 """
 Hillary Elrick, September 16th, 2019
@@ -548,15 +548,14 @@ def performGuideSearch(self):
         batchID = binascii.b2a_hex(os.urandom(9)).decode('utf-8')
 
         genome_fa = os.path.join(self.dbConnection.ROOT_PATH,'jbrowse', 'data', self.genome,"processed",self.genome+".fa")
-        twoBitToFa_path = os.path.join(self.dbConnection.ROOT_PATH,'bin/twoBitToFa')
         genome_2bit = os.path.join(self.dbConnection.ROOT_PATH,'jbrowse', 'data', self.genome,"processed",self.genome+'.2bit')
         tempfiles_path = os.path.join(self.dbConnection.ROOT_PATH,'src/guide-finder/tempfiles')
         if self.cli:
             import time
             time_0 = time.time()
             print("Fetching sequence...")
 
-        get_sequence.fetch_sequence(twoBitToFa_path, self.searchInput, genome_2bit, os.path.join(tempfiles_path,batchID+'_out.fa'))
+        get_sequence.fetch_sequence(self.searchInput, genome_2bit, os.path.join(tempfiles_path,batchID+'_out.fa'))
         if self.cli:
             time_1 = time.time()
             print("Finished fetching sequence. " + str(round(time_1-time_0,4)))

Diff for: docker-compose.yaml

@@ -296,7 +296,7 @@ def convertExtendedBedToFasta(batchID, genome, genome_fa, tempfile_directory):
 		raise Exception("Error in bedtools getfasta command: " + str(p.stderr))
 
 	"""
-	twoBitTwoFaCommand = [os.path.join(dir_path,"../../../bin/twoBitToFa"), os.path.join(dir_path,"../../../jbrowse/data/mm10/processed/mm10.2bit"),"-bed="+extendedBed, extendedFasta]
+	twoBitTwoFaCommand = ["twoBitToFa", os.path.join(dir_path,"../../../jbrowse/data/mm10/processed/mm10.2bit"),"-bed="+extendedBed, extendedFasta]
 	p = run(twoBitTwoFaCommand,stderr=PIPE)
 	if p.stderr:
 		raise Exception("Error in twoBitToFa: "+str(p.stderr))

Diff for: src/guide-finder/GuideSearchAndScore.py

@@ -1,3 +1,5 @@
+#!/usr/bin/env python3
+
 import subprocess
 import sys
 import os
@@ -12,31 +14,30 @@
 
 dir_path = os.path.dirname(os.path.abspath(__file__))
 
-def fetch_sequence(twobit_to_fa, chrom_coord, genome_twobit, output_fasta):
+def fetch_sequence(chrom_coord, genome_twobit, output_fasta):
     '''
-        This function will return sequence for a given chromosome, start, end from the genome_twobit file	
-	
+        This function will return sequence for a given chromosome, start, end from the genome_twobit file
     '''
 
     chrom_match = re.match(r"(.+):(\d+)-(\d+)", chrom_coord)
     if chrom_match is None:
         sys.exit("Chromosomal coordinate must be in the format chrom:start-end")
-    
+
     chrom, start, end = chrom_match.group(1), chrom_match.group(2), chrom_match.group(3)
     # end for twoBitToFa is non-inclusive, add extra base
     start = str(int(start)-1)
     seq_param, start_param, end_param = '-seq={0}'.format(chrom), '-start={0}'.format(start), '-end={0}'.format(end)
     tmp_fasta = os.path.join(dir_path,"tmp.fa")
 
     try:
-        subprocess.run([twobit_to_fa, seq_param, start_param, end_param, genome_twobit, tmp_fasta])  
+        subprocess.run(["twoBitToFa", seq_param, start_param, end_param, genome_twobit, tmp_fasta])
     except Exception as err:
         print(err)
 
     if not os.path.isfile(tmp_fasta) or os.path.getsize(tmp_fasta) == 0:
         print("twobitToFasta failed: "+" ".join([twobit_to_fa, seq_param, start_param, end_param, genome_twobit, tmp_fasta]))
         sys.exit("sequence is empty for given parameters. Please check your parameters again")
-    
+
     with open(tmp_fasta,"r") as inp_fa, open(output_fasta,"w") as out_fa:
         out_fa.write(">"+chrom+":"+str(int(start)+1)+"-"+end+"\n")
         for line in inp_fa:
@@ -49,15 +50,8 @@ def fetch_sequence(twobit_to_fa, chrom_coord, genome_twobit, output_fasta):
     return 1
 
 
-
-#main block 
 if __name__ == "__main__":
-
-    if len(sys.argv) !=5:
-        sys.exit("Need path to twoBittoFa executable, chromosomal coordinates (chr:start-stop),  genome_twobit, output fasta file file as arguments.")
-	
-    twobit_to_fa, chrom_coord, genome_twobit, output_fasta = sys.argv[1:]
-    fetch_sequence(twobit_to_fa, chrom_coord, genome_twobit, output_fasta)
-
-
-
+    if len(sys.argv) != 5:
+        sys.exit("Need chromosomal coordinates (chr:start-stop), genome_twobit, output fasta file as arguments.")
+    chrom_coord, genome_twobit, output_fasta = sys.argv[1:]
+    fetch_sequence(chrom_coord, genome_twobit, output_fasta)

Diff for: src/guide-finder/core/find_offtargets.py

@@ -5,7 +5,6 @@ set -euo pipefail
 ASSEMBLY="$1"
 
 SETUP_BIN=$(dirname $(realpath "$0"))
-PATH="/var/www/html/bin:/opt/dicey/bin:$PATH"
 
 gunzip --keep *.fa.gz *.gff*.gz
 FASTA=$(echo *.fa) # only one