Run bigscape v2

bin/install-nplinker-deps

@@ -125,6 +125,7 @@ pip install -q -U pip setuptools
 echo "🔥 Start installing BigScape ..."
     [[ -d BiG-SCAPE ]] || git clone https://github.com/medema-group/BiG-SCAPE.git
     cd BiG-SCAPE
+    git reset --hard
     git config --add advice.detachedHead false  # disable advice
     git config pull.ff only
     git checkout master
@@ -136,6 +137,21 @@ echo "🔥 Start installing BigScape ..."
     chmod 775 Annotated_MIBiG_reference
     ln -sf $LIB_PATH/BiG-SCAPE/bigscape.py $PY_PATH/bin
     cd ..
+    # blob size limit to remove large files left in history
+    [[ -d BiG-SCAPE-v2 ]] || git clone -b dev --filter=blob:limit=10m https://github.com/medema-group/BiG-SCAPE.git BiG-SCAPE-v2
+    cd BiG-SCAPE-v2
+    git config --ad advice.detatchedHead false
+    git checkout dfb0d78427e020aab2c72cc741327ccd102470a1 # specific commit that includes important fixes to v2
+    pip install click
+    pip install sqlalchemy
+    pip install pyhmmer
+    chmod 754 bigscape.py
+    ln -sf $LIB_PATH/BiG-SCAPE-v2/bigscape.py $PY_PATH/bin/bigscape-v2.py
+    ln -sf $LIB_PATH/BiG-SCAPE-v2/config.ini $PY_PATH/bin # new system of configuration needs default config file
+    ln -sf $LIB_PATH/BiG-SCAPE-v2/big_scape $PY_PATH/bin # folder is needed for some files
+    cd ..
+
+
 echo -e "✅ BigScape installed successfully\n"
 
 #--- Install FastTree (not support Windows, required by BigScape)

src/nplinker/arranger.py

@@ -263,21 +263,34 @@ def _run_bigscape(self) -> None:
         default BiG-SCAPE directory.
         """
         defaults.BIGSCAPE_RUNNING_OUTPUT_PATH.mkdir(exist_ok=True, parents=True)
+
+        version = config.bigscape.version
+
         run_bigscape(
             defaults.ANTISMASH_DEFAULT_PATH,
             defaults.BIGSCAPE_RUNNING_OUTPUT_PATH,
             config.bigscape.parameters,
+            version,
         )
-        for f in glob(
-            str(
-                defaults.BIGSCAPE_RUNNING_OUTPUT_PATH
-                / "network_files"
-                / "*"
-                / "mix"
-                / "mix_clustering_c*.tsv"
+
+        if version == 1:
+            for f in glob(
+                str(
+                    defaults.BIGSCAPE_RUNNING_OUTPUT_PATH
+                    / "network_files"
+                    / "*"
+                    / "mix"
+                    / "mix_clustering_c*.tsv"
+                )
+            ):
+                shutil.copy(f, defaults.BIGSCAPE_DEFAULT_PATH)
+        elif version == 2:
+            shutil.copy(
+                defaults.BIGSCAPE_RUNNING_OUTPUT_PATH / "data_sqlite.db",
+                defaults.BIGSCAPE_DEFAULT_PATH,
             )
-        ):
-            shutil.copy(f, defaults.BIGSCAPE_DEFAULT_PATH)
+        else:
+            raise ValueError(f"Invalid BiG-SCAPE version: {version}")
 
     def arrange_strain_mappings(self) -> None:
         """Arrange the strain mappings file.
@@ -319,7 +332,9 @@ def _validate_strain_mappings(self) -> None:
 
     def _generate_strain_mappings(self) -> None:
         """Generate the strain mappings file for the PODP mode."""
-        podp_json_file = defaults.DOWNLOADS_DEFAULT_PATH / f"paired_datarecord_{config.podp_id}.json"
+        podp_json_file = (
+            defaults.DOWNLOADS_DEFAULT_PATH / f"paired_datarecord_{config.podp_id}.json"
+        )
         genome_status_json_file = defaults.DOWNLOADS_DEFAULT_PATH / GENOME_STATUS_FILENAME
         genome_bgc_mappings_file = defaults.ANTISMASH_DEFAULT_PATH / GENOME_BGC_MAPPINGS_FILENAME
         gnps_file_mapping_file = self.gnps_file_mappings_file

src/nplinker/genomics/bigscape/runbigscape.py

@@ -15,22 +15,46 @@ def run_bigscape(
     antismash_path: str | PathLike,
     output_path: str | PathLike,
     extra_params: str,
+    version: int = 1,
 ):
-    bigscape_py_path = "bigscape.py"
+    """Runs BiG-SCAPE to cluster BGCs.
+
+    The behavior of this function is slightly different depending on the version of
+    BiG-SCAPE that is set to run using the configuration file.
+    Mostly this means a different set of parameters is used between the two versions.
+    """
+    # switch to correct version of BiG-SCAPE
+    if version == 1:
+        bigscape_py_path = "bigscape.py"
+    elif version == 2:
+        bigscape_py_path = "bigscape-v2.py"
+
     logger.info(
         f'run_bigscape: input="{antismash_path}", output="{output_path}", extra_params={extra_params}"'
     )
 
     try:
         subprocess.run([bigscape_py_path, "-h"], capture_output=True, check=True)
     except Exception as e:
-        raise Exception(f"Failed to find/run bigscape.py (path={bigscape_py_path}, err={e})") from e
+        raise Exception(
+            f"Failed to find/run bigscape.py (path={bigscape_py_path}, err={e})"
+        ) from e
 
     if not os.path.exists(antismash_path):
         raise Exception(f'antismash_path "{antismash_path}" does not exist!')
 
-    # configure the IO-related parameters, including pfam_dir
-    args = [bigscape_py_path, "-i", antismash_path, "-o", output_path, "--pfam_dir", PFAM_PATH]
+    # assemble arguments. first argument is the python file
+    args = [bigscape_py_path]
+
+    # version 2 points to specific Pfam file, version 1 points to directory
+    # version 2 also requires the cluster subcommand
+    if version == 1:
+        args.extend(["--pfam_dir", PFAM_PATH])
+    elif version == 2:
+        args.extend(["cluster", "--pfam_path", PFAM_PATH + "/Pfam-A.hmm"])
+
+    # add input and output paths. these are unchanged
+    args.extend(["-i", str(antismash_path), "-o", str(output_path)])
 
     # append the user supplied params, if any
     if len(extra_params) > 0:

src/nplinker/nplinker_default.toml

@@ -9,7 +9,12 @@ to_use = true
 version = "3.1"
 
 [bigscape]
+verison = 1
 parameters = "--mibig --clans-off --mix --include_singletons --cutoffs 0.30"
+
+# for version 2, use the following parameters string:
+# parameters = "--mibig_version 3.1 --include_singletons --gcf_cutoffs 0.30"
+
 cutoff = "0.30"
 
 [scoring]