kssdtree 1.1.9__tar.gz → 2.0.1__tar.gz

{kssdtree-1.1.9/kssdtree.egg-info → kssdtree-2.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kssdtree
-Version: 1.1.9
+Version: 2.0.1
 Summary: Kssdtree is a versatile Python package for phylogenetic analysis. It also provides one-stop tree construction and visualization. It can handle DNA sequences of both fasta or fastq format, whether gzipped or not. 
 Home-page: https://github.com/yhlink/kssdtree
 Download-URL: https://pypi.org/project/kssdtree

{kssdtree-1.1.9 → kssdtree-2.0.1/kssdtree.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kssdtree
-Version: 1.1.9
+Version: 2.0.1
 Summary: Kssdtree is a versatile Python package for phylogenetic analysis. It also provides one-stop tree construction and visualization. It can handle DNA sequences of both fasta or fastq format, whether gzipped or not. 
 Home-page: https://github.com/yhlink/kssdtree
 Download-URL: https://pypi.org/project/kssdtree

{kssdtree-1.1.9 → kssdtree-2.0.1}/kssdtree.py

@@ -26,12 +26,15 @@ def sketch(shuf_file=None, genome_files=None, output=None, set_opt=None):
                     print('Genome format error for file:', filename)
                     return False
         if not os.path.exists(shuf_file):
-            if shuf_file in ['L3K9.shuf', './L3K9.shuf']:
+            if shuf_file in ['L3K9.shuf', './L3K9.shuf', 'L3K10.shuf', './L3K10.shuf']:
                 print('Downloading...', shuf_file)
                 import http.client
                 http.client.HTTPConnection._http_vsn = 10
                 http.client.HTTPConnection._http_vsn_str = 'HTTP/1.0'
-                url = 'http://www.metakssdcoabundance.link/kssdtree/shuffle/' + shuf_file
+                if shuf_file == 'L3K9.shuf' or shuf_file == './L3K9.shuf':
+                    url = 'https://zenodo.org/records/12699159/files/L3K9.shuf?download=1'
+                else:
+                    url = 'https://zenodo.org/records/12699159/files/L3K10.shuf?download=1'
                 start_time = time.time()
                 response = requests.get(url, stream=True)
                 with open(shuf_file, 'wb') as file:
@@ -44,8 +47,7 @@ def sketch(shuf_file=None, genome_files=None, output=None, set_opt=None):
                         "Network timeout, please manually download from github (https://github.com/yhlink/kssdtree/tree/master/shuffle_file)")
                     return False
                 print('Download finished: ', shuf_file)
-            elif shuf_file in ['L2K8.shuf', 'L3K10.shuf', 'L2K9.shuf', 'L3K11.shuf', './L2K8.shuf', './L3K10.shuf',
-                               './L2K9.shuf', './L3K11.shuf']:
+            elif shuf_file in ['L2K8.shuf', 'L2K9.shuf', 'L3K11.shuf', './L2K8.shuf', './L2K9.shuf', './L3K11.shuf']:
                 print('Shuffling...', shuf_file)
                 file_name = shuf_file.split('.')[0]
                 k = int(file_name[3:])
@@ -74,14 +76,14 @@ def sketch(shuf_file=None, genome_files=None, output=None, set_opt=None):
         return False
 
 
-def dist(ref_sketch=None, qry_sketch=None, output=None, flag=None):
-    if ref_sketch is not None and qry_sketch is not None and output is not None:
-        if not os.path.exists(ref_sketch):
-            print('No such file or directory: ', ref_sketch)
-            return False
-        if not os.path.exists(qry_sketch):
-            print('No such file or directory: ', qry_sketch)
+def dist(genome_sketch=None, output=None, flag=None):
+    if genome_sketch is not None and output is not None:
+        if not os.path.exists(genome_sketch):
+            print('No such file or directory: ', genome_sketch)
             return False
+        # if not os.path.exists(qry_sketch):
+        #     print('No such file or directory: ', qry_sketch)
+        #     return False
         if flag is None:
             flag = 0
         print('Disting...')
@@ -95,7 +97,7 @@ def dist(ref_sketch=None, qry_sketch=None, output=None, flag=None):
         else:
             output_name = output
         if output_name.endswith(".phy") or output_name.endswith(".phylip"):
-            kssd.dist_dispatch(ref_sketch, output, qry_sketch, 2, 0, flag)
+            kssd.dist_dispatch(genome_sketch, output, genome_sketch, 2, 0, flag)
             end = time.time()
             print('Dist spend time：%.2fs' % (end - start))
             print('Dist finished!')
@@ -108,23 +110,26 @@ def dist(ref_sketch=None, qry_sketch=None, output=None, flag=None):
         return False
 
 
-def retrieve(ref_sketch=None, qry_sketch=None, output=None, N=None, method=None):
-    if ref_sketch is not None and qry_sketch is not None and output is not None:
+def retrieve(database=None, genome_sketch=None, output=None, N=None, method=None):
+    if database is not None and genome_sketch is not None and output is not None:
         if method is None:
             method = 'nj'
         if method not in ['nj', 'dnj']:
             print('Only support nj and dnj methods!!!')
             return
-        if not os.path.exists(qry_sketch):
-            print('No such file or directory: ', qry_sketch)
+        if not os.path.exists(genome_sketch):
+            print('No such file or directory: ', genome_sketch)
             return False
-        if ref_sketch == 'gtdbr214_sketch':
+        if database == 'gtdbr214':
             print('Retrieving...')
             start = time.time()
             if not os.path.exists(output):
                 os.makedirs(output)
                 print("Created directory:", output)
-            newick, accession_taxonomy = toolutils.upload_request(qry_sketch=qry_sketch, method=method, N=N)
+            else:
+                print('Output path exist!!!')
+                return False
+            newick, accession_taxonomy = toolutils.upload_request(qry_sketch=genome_sketch, method=method, N=N)
             if newick is None:
                 print('Server error!!!')
                 return False
@@ -138,7 +143,7 @@ def retrieve(ref_sketch=None, qry_sketch=None, output=None, N=None, method=None)
             print('Retrieve finished!')
             return True
         else:
-            print("ref_sketch must be set to 'gtdbr214_sketch'")
+            print("database only support 'gtdbr214'")
             return False
     else:
         print('Args error!!!')
@@ -184,6 +189,12 @@ def build(phylip=None, output=None, method=None):
                 print('Build spend time：%.2fs' % (end - start))
                 print('Build finished!')
                 return True
+            else:
+                print('phylip format error, Check that the phylip format is consistent with NJ or DNJ requirements!!!')
+                return False
+        else:
+            print('Output type error, only supports .newick (.nwk) format:', output_name)
+            return False
     else:
         print('Args error!!!')
         return False
@@ -266,8 +277,8 @@ def subtract(ref_sketch=None, genome_sketch=None, output=None, flag=None):
         return False
 
 
-def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxonomy=None, method='nj', mode='r', N=0):
-    if reference is None and taxonomy is None:
+def quick(shuf_file=None, genome_files=None, output=None, reference=None, database=None, method='nj', mode='r', N=0):
+    if reference is None and database is None:
         if shuf_file is not None and genome_files is not None and output is not None:
             timeStamp = int(time.mktime(time.localtime(time.time())))
             temp_sketch = toolutils.rs() + '_sketch_' + str(timeStamp)
@@ -286,9 +297,9 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
                 return False
             print('Step2...')
             if method == 'nj':
-                s2 = dist(ref_sketch=temp_sketch, qry_sketch=temp_sketch, output=temp_phy, flag=0)
+                s2 = dist(genome_sketch=temp_sketch, output=temp_phy, flag=0)
             else:
-                s2 = dist(ref_sketch=temp_sketch, qry_sketch=temp_sketch, output=temp_phy, flag=1)
+                s2 = dist(genome_sketch=temp_sketch, output=temp_phy, flag=1)
             if not s2:
                 return False
             print('Step3...')
@@ -297,7 +308,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
                 return False
             print('Step4...')
             print('Tree visualization finished!')
-            visualize(newick=output, taxonomy=taxonomy, mode=mode)
+            visualize(newick=output, mode=mode)
             if platform.system() == 'Linux':
                 current_directory = os.getcwd()
                 temp_dir1 = os.path.join(current_directory, temp_sketch)
@@ -311,8 +322,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
         else:
             print('Args error, please see https://kssdtree.readthedocs.io/en/latest!!!')
             return False
-
-    elif reference == 'gtdbr214_sketch' and taxonomy is None:
+    elif reference is None and database == 'gtdbr214':
         if shuf_file is not None and genome_files is not None and output is not None:
             if not toolutils.is_positive_integer(N):
                 print("N must >0 !!!")
@@ -325,7 +335,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
             s1 = sketch(shuf_file=shuf_file, genome_files=genome_files, output=qry_sketch, set_opt=True)
             if not s1:
                 return False
-            s2 = retrieve(ref_sketch=reference, qry_sketch=qry_sketch, output=output, N=N, method=method)
+            s2 = retrieve(database=database, genome_sketch=qry_sketch, output=output, N=N, method=method)
             if not s2:
                 return False
             print('Tree visualization finished!')
@@ -342,7 +352,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
         else:
             print('Args error, please see https://kssdtree.readthedocs.io/en/latest!!!')
             return False
-    else:
+    elif reference is not None and database is None:
         if shuf_file is not None and genome_files is not None and output is not None and method in ['nj', 'dnj']:
             if shuf_file is not None and genome_files is not None and output is not None and method in ['nj', 'dnj']:
                 if not toolutils.allowed_file(genome_files):
@@ -367,7 +377,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
                 else:
                     temp_union_sketch = temp_reference_sketch
                 temp_subtract_sketch = toolutils.rs() + '_subtract_sketch_' + str(timeStamp)
-                temp_phy = toolutils.rs() +'_temp.phy'
+                temp_phy = toolutils.rs() + '_temp.phy'
                 print('Step1...')
                 s1 = sketch(shuf_file=shuf_file, genome_files=reference, output=temp_reference_sketch, set_opt=True)
                 if not s1:
@@ -385,10 +395,10 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
                     return False
                 print('Step3...')
                 if method == 'nj':
-                    s5 = dist(ref_sketch=temp_subtract_sketch, qry_sketch=temp_subtract_sketch, output=temp_phy,
+                    s5 = dist(genome_sketch=temp_subtract_sketch, output=temp_phy,
                               flag=0)
                 else:
-                    s5 = dist(ref_sketch=temp_subtract_sketch, qry_sketch=temp_subtract_sketch, output=temp_phy,
+                    s5 = dist(genome_sketch=temp_subtract_sketch, output=temp_phy,
                               flag=1)
                 if not s5:
                     return False
@@ -398,7 +408,7 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
                     return False
                 print('Step5...')
                 print('Tree visualization finished!')
-                visualize(newick=output, taxonomy=taxonomy, mode=mode)
+                visualize(newick=output, mode=mode)
                 if platform.system() == 'Linux':
                     current_directory = os.getcwd()
                     temp_dir1 = os.path.join(current_directory, temp_reference_sketch)
@@ -421,3 +431,6 @@ def quick(shuf_file=None, genome_files=None, output=None, reference=None, taxono
             else:
                 print('Args error, please see https://kssdtree.readthedocs.io/en/latest!!!')
                 return False
+    else:
+        print('Pipeline error, please see https://kssdtree.readthedocs.io/en/latest!!!')
+        return False

{kssdtree-1.1.9 → kssdtree-2.0.1}/setup.py

@@ -1,37 +1,41 @@
 import sys
-
 from setuptools import setup, Extension, find_packages
 from os import environ
-import subprocess
 import os
+import platform
 
 
-def get_gcc_version():
-    gcc_version = subprocess.check_output(['gcc', '--version']).decode('utf-8')
-    version_line = gcc_version.split('\n', 1)[0]
-    version_str = version_line.split()[-1]
-    return version_str
-
+# def get_gcc_version():
+#     gcc_version = subprocess.check_output(['gcc', '--version']).decode('utf-8')
+#     version_line = gcc_version.split('\n', 1)[0]
+#     version_str = version_line.split()[-1]
+#     return version_str
+extra_compile_args = []
+extra_link_args = []
 if 'darwin' in sys.platform:
-    gcc_version = get_gcc_version()
-    if '9.' in gcc_version:
-        gcc_path = "/opt/homebrew/bin/gcc-9"
-    elif '10.' in gcc_version:
-        gcc_path = "/opt/homebrew/bin/gcc-10"
-    elif '11.' in gcc_version:
-        gcc_path = "/opt/homebrew/bin/gcc-11"
-    elif '12.' in gcc_version:
-        gcc_path = "/opt/homebrew/bin/gcc-12"
-    elif '14.' in gcc_version:
-        gcc_path = "/opt/homebrew/bin/gcc-14"
-    else:
-        gcc_path = "/opt/homebrew/bin/gcc-13"
-    gcc_command = 'gcc-' + str(gcc_version)
-    extra_compile_args = ['-fopenmp']
-    extra_link_args = ['-fopenmp']
-    os.environ["CC"] = gcc_path
+    target_dirs = ["gcc-9", "gcc-10", "gcc-11", "gcc-12", "gcc-13", "gcc-14", "gcc-15"]
+    path = "/opt/homebrew/bin/"
+    directories = [dir for dir in os.listdir(path) if dir in target_dirs]
+    if len(directories) > 0:
+        gcc_version = directories[0]
+        if 'gcc-9' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-9"
+        elif 'gcc-10' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-10"
+        elif 'gcc-11' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-11"
+        elif 'gcc-12' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-12"
+        elif 'gcc-14' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-14"
+        elif 'gcc-15' == gcc_version:
+            gcc_path = "/opt/homebrew/bin/gcc-15"
+        else:
+            gcc_path = "/opt/homebrew/bin/gcc-13"
+        extra_compile_args = ['-fopenmp']
+        extra_link_args = ['-fopenmp']
+        os.environ["CC"] = gcc_path
 else:
-    gcc_command = 'gcc'
     if environ.get('CC') and 'clang' in environ['CC']:
         # clang
         extra_compile_args = ['-fopenmp=libomp']
@@ -81,16 +85,25 @@ include_dirs1 = ['kssdheaders']
 include_dirs2 = ['njheaders']
 include_dirs3 = ['dnjheaders']
 
-require_pakages = [
-    'pyqt5',
-    'ete3',
-    'requests'
-]
-
+if 'darwin' in sys.platform:
+    if platform.machine() == 'arm64':
+        require_pakages = []
+    else:
+        require_pakages = [
+            'pyqt5',
+            'ete3',
+            'requests'
+        ]
+else:
+    require_pakages = [
+        'pyqt5',
+        'ete3',
+        'requests'
+    ]
 
 setup(
     name='kssdtree',
-    version='1.1.9',
+    version='2.0.1',
     author='Hang Yang',
     author_email='yhlink1207@gmail.com',
     description="Kssdtree is a versatile Python package for phylogenetic analysis. It also provides one-stop tree construction and visualization. It can handle DNA sequences of both fasta or fastq format, whether gzipped or not. ",