Warwick

modules/assemble.py

@@ -385,7 +385,9 @@ def do_spades(self, reads) :
         if os.path.isdir(outdir) :
             shutil.rmtree(outdir)
         read_len = min(self.__get_read_len(reads), 140)
-        kmer = ','.join([str(max(min(int(read_len*float(x)/200)*2+1, 127),17)) for x in parameters['kmers'].split(',')])
+        kmer = [str(max(min(int(read_len * float(x) / 200) * 2 + 1, 127), 17)) for x in parameters['kmers'].split(',')]
+        # Ensure there are no duplicate kmers
+        kmer = ','.join(list(dict.fromkeys(kmer)))
         read_input = []
         for lib_id, lib in enumerate(reads) :
             if len(lib) == 1 :
@@ -711,9 +713,18 @@ def __readAssembly(self, assembly) :
         return seq, fasfile
 
     def do_kraken(self, assembly, seq) :
+        window_size = 10000
         with open(assembly+'.filter', 'w') as fout :
+            #  Kraken calculates species percentages based on the number of contigs
+            #  so large contigs, such as found with long read assemblies are not
+            #  appropriately represented
             for n, s in sorted(seq.items()) :
-                if s[0] > 1000 :
+                if s[0] > (2 * window_size):
+                    N = s[0]//window_size
+                    size = s[0]//N
+                    for i in range(N):
+                        fout.write('>{0}_{1}\n{2}\n'.format(n, i, s[2][i*size:((i+1)*size)-1]))
+                elif s[0] > 1000 :
                     fout.write('>{0}\n{1}\n'.format(n, s[2]))
         cmd = '{kraken2} -db {kraken_database} --threads {n_cpu} --output - --report {assembly}.kraken {assembly}.filter'.format(
             assembly=assembly, **parameters

modules/configure.py

@@ -216,7 +216,7 @@ def download_krakenDB() :
         os.makedirs('minikraken2')
     os.chdir(os.path.join(moveTo, 'minikraken2'))
     if not os.path.exists('hash.k2d') or  not os.path.getsize('hash.k2d') == 8000000032:
-        minikraken_url = 'ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/old/minikraken2_v2_8GB_201904.tgz'
+        minikraken_url = 'https://genome-idx.s3.amazonaws.com/kraken/minikraken2_v2_8GB_201904.tgz'
         logger('Downloading minikraken2 from {0}. This might take a long time.'.format(minikraken_url))
         subprocess.Popen('curl -Lo minikraken2_v2_8GB.tgz {0}'.format(minikraken_url).split(), stderr=subprocess.PIPE).communicate()
         logger('Unpackaging minikraken2.')
@@ -261,7 +261,7 @@ def install_externals() :
     if not getExecutable([externals['hapog']]) :
 
         # HAPO depends on installed htslib which may not be available on host, so do a local install
-        htslib_ver = '1.3.2'
+        htslib_ver = '1.15.1'
         url = 'https://github.com/samtools/htslib/releases/download/{0}/htslib-{0}.tar.bz2'.format(htslib_ver)
         logger('Downloading htslib from {0}'.format(url))
         subprocess.Popen('curl -Lo htslib.tar.bz2 {0}'.format(url).split(), stderr=subprocess.PIPE).communicate()
@@ -276,7 +276,7 @@ def install_externals() :
         shutil.rmtree('htslib-{0}'.format(htslib_ver))
         logger('Done\n')
 
-        hapog_ver = '1.2'
+        hapog_ver = '1.3.4'
         hapog_name = 'HAPO-G-'+hapog_ver
         url = 'https://github.com/institut-de-genomique/HAPO-G/archive/refs/tags/{0}.tar.gz'.format(hapog_ver)
         logger('Downloading Hapo-G package from {0}'.format(url))
@@ -285,35 +285,23 @@ def install_externals() :
         subprocess.Popen('tar -xzf hapog.tar.gz'.split()).communicate()
 
         os.unlink('hapog.tar.gz')
-        gcc_ver = int(subprocess.Popen(['gcc', '-dumpversion'], stdout=subprocess.PIPE).communicate()[0].
-                      strip().split(b'.')[0])
-        if gcc_ver < 9:
-            #  Compiling hapog with earlier versions of gcc creates an executable that runs
-            #  but fails with a segmentation fault when processing data
-            #  Use precompiled which works with a locally compileed htslib
-            os.makedirs('HAPO-G-{0}/build'.format(hapog_ver), exist_ok=True)
-            shutil.copy('../bin/hapog','HAPO-G-1.2/build')
-            os.makedirs('HAPO-G-{0}/bin'.format(hapog_ver), exist_ok=True)
-            subprocess.Popen('ln -fs ../build/hapog HAPO-G-{0}/bin/hapog'.format(hapog_ver).
-                             split(), stderr=subprocess.PIPE).communicate()
-        else:
-            my_env = os.environ.copy()
-
-            if not getExecutable('cmake') or int(os.popen('cmake --version').read().split()[2].split('.')[0]) < 3:
-                url = 'https://github.com/Kitware/CMake/releases/download/v3.25.1/cmake-3.25.1-linux-x86_64.tar.gz'
-                logger('Installing cmake from {0} for hapog installation'.format(url))
-                subprocess.Popen('curl -Lo cmake.tar.gz {0}'.format(url).split(), stderr=subprocess.PIPE).communicate()
-                logger('Unpackaging cmake package')
-                subprocess.Popen('tar -xzf cmake.tar.gz'.split()).communicate()
-                subprocess.Popen('ln -fs cmake-3.25.1-linux-x86_64/bin/cmake .'.split(),
-                                 stderr=subprocess.PIPE).communicate()
-                os.unlink('cmake.tar.gz')
-                my_env["PATH"] = externals_dir + ':' + my_env["PATH"]
-
-            subprocess.Popen('bash build.sh -l {0}'.format(htslib_dir), shell=True, stdout=subprocess.PIPE,
-                             stderr=subprocess.PIPE, cwd='HAPO-G-{0}'.format(hapog_ver),  env=my_env).communicate()
-
-        subprocess.Popen('ln -fs HAPO-G-{0}/hapog.py ./hapog.py'.format(hapog_ver).split(),
+        my_env = os.environ.copy()
+
+        if not getExecutable('cmake') or int(os.popen('cmake --version').read().split()[2].split('.')[0]) < 3:
+            url = 'https://github.com/Kitware/CMake/releases/download/v3.25.1/cmake-3.25.1-linux-x86_64.tar.gz'
+            logger('Installing cmake from {0} for hapog installation'.format(url))
+            subprocess.Popen('curl -Lo cmake.tar.gz {0}'.format(url).split(), stderr=subprocess.PIPE).communicate()
+            logger('Unpackaging cmake package')
+            subprocess.Popen('tar -xzf cmake.tar.gz'.split()).communicate()
+            subprocess.Popen('ln -fs cmake-3.25.1-linux-x86_64/bin/cmake .'.split(),
+                             stderr=subprocess.PIPE).communicate()
+            os.unlink('cmake.tar.gz')
+            my_env["PATH"] = externals_dir + ':' + my_env["PATH"]
+
+        subprocess.Popen('bash build.sh -l {0}'.format(htslib_dir), shell=True, stdout=subprocess.PIPE,
+                         stderr=subprocess.PIPE, cwd='HAPO-G-{0}'.format(hapog_ver),  env=my_env).communicate()
+
+        subprocess.Popen('ln -s HAPO-G-{0}/hapog.py ./hapog.py'.format(hapog_ver).split(),
                          stderr=subprocess.PIPE).communicate()
         subprocess.Popen('chmod 755 ./hapog.py'.split(), stderr=subprocess.PIPE).communicate()
 
@@ -475,7 +463,7 @@ def install_externals() :
         logger('Done\n')
 
     if not getExecutable([externals['blastn']]) or not getExecutable([externals['makeblastdb']]) :
-        blast_url = 'ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.8.1/ncbi-blast-2.8.1+-x64-linux.tar.gz'
+        blast_url = 'https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.8.1/ncbi-blast-2.8.1+-x64-linux.tar.gz'
         logger('Downloading ncbi-blast package from {0}'.format(blast_url))
         subprocess.Popen('curl -Lo ncbi-blast-2.8.1+-x64-linux.tar.gz {0}'.format(blast_url).split(), stderr=subprocess.PIPE).communicate()
         logger('Unpackaging ncbi-blast package'.format(blast_url))